Polarimetric Synthetic Aperture Radar (PolSAR) image classification remains challenging due to complex scattering mechanisms, speckle noise, and the difficulty of selecting the most informative features from the multitude of derivable polarimetric representations. To address these challenges, we propose a novel two-step methodology for PolSAR image classification that utilizes an advanced feature extraction technique while leveraging the strengths of a vision transformer(ViT) architecture. In the first step, advanced feature extraction techniques are explored to capture multiscale spatial scattering information and preserve crucial structural information of the PolSAR image while effectively mitigating the noise present on it. In the second step, a dual-branch ViT (DB-ViT) is proposed that simultaneously processes both the original polarimetric features and the extracted spatial features, enabling effective information fusion through a local window attention transformer (LWAT). Extensive experiments on the Flevoland AIRSAR and the San-Francisco RADARSAT-2 benchmark datasets demonstrated that our approach consistently outperforms state-of-the-art methods, achieving the highest overall accuracies of 99.50% and 99.51%, respectively.

Fine-grained classification is one among the major research hotspots in remote sensing (RS) image interpretation that has gained attention span for several applications. RS image classification at the finest level is a highly challenging task to perform as it should consider the presence of high intra-class similarity, subtle inter-class differences and spatial scale variations. To address this issue, we propose an end-to-end saliency-guided feature mining network (SGFM-Net) specifically crafted to extract essential features that enable differentiation of highly similar instances. The essential components in the proposed SGFM-Net framework includes: i) a modified ConvNeXt as the backbone architecture that integrates pyramid convolutions to extract the multiscale features; ii) a saliency-guided feature mining module; iii) a multiple hierarchies attention module (MHAM) to filter and enhance intrinsic features; iv) a bilinear polymerization pooling (BPP) to fuse intrinsic and attention maps into a discriminative vector and v) a feature mapping network (FMN) to decouple and normalize that vector before classification. Experimental results carried out on three benchmark archives: FGSC-23, Aircraft-16 and FGSCR-42 have shown remarkable improvements over existing state-of-the-art methods.

Semantic segmentation of remote sensing imagery (RSI) is a fundamental task that aims at assigning a category label to each pixel. To pursue precise segmentation with one or more fine-grained categories, semantic segmentation often requires holistic segmentation of whole-scene RSI (WRI), which is normally characterized by a large size. However, conventional deep learning methods struggle to handle holistic segmentation of WRI due to the memory limitations of the graphics processing unit (GPU), thus requiring to adopt suboptimal strategies such as cropping or fusion, which result in performance degradation. Here, we introduce the Robust End-to-end semantic Segmentation architecture for whole-scene remoTe sensing imagery (REST). REST is the first intrinsically endtoend framework for truly holistic segmentation of WRI, supporting a wide range of encoders and decoders in a plugandplay fashion. It enables seamless integration with mainstream semantic segmentation methods, and even more advanced foundation models. Specifically, we propose a novel spatial parallel interaction mechanism (SPIM) within REST to overcome GPU memory constraints and achieve global context awareness. Unlike traditional parallel methods, SPIM enables REST to process a WRI effectively and efficiently by combining parallel computation with a divideandconquer strategy. Both theoretical analysis and experiments demonstrate that REST attains nearlinear throughput scalability as additional GPUs are employed. Extensive experiments demonstrate that REST consistently outperforms existing cropping-based and fusion-based methods across a variety of scenarios, ranging from single-class to multi-class segmentation, from multispectral to hyperspectral imagery, and from satellite to drone platforms. The robustness and versatility of REST are expected to offer a promising solution for the holistic segmentation of WRI, with the potential for further extension to large-size medical imagery segmentation.

Deep learning-based methods have achieved significant success in remote sensing Earth observation data analysis. Numerous feature fusion techniques address multimodal remote sensing image classification by integrating global and local features. However, these techniques often struggle to extract structural and detail features from heterogeneous and redundant multimodal images, particularly in label-scarce scenarios. With the goal of introducing frequency domain learning to model key and sparse detail features, this paper introduces the spatial–spectral-frequency interaction network (S 2 Fin), which integrates pairwise fusion modules across the spatial, spectral, and frequency domains. Specifically, we propose a high-frequency sparse enhancement transformer to refine spectral signatures by adaptively enhancing discriminative high-frequency components. For spatial-frequency interaction, we present a depth-wise strategy: the adaptive frequency channel module fuses low-frequency structural information with enhanced details in shallow layers, while the high-frequency resonance mask amplifies modality-consistent regions in deep layers using phase similarity. In addition, a spatial–spectral attention fusion module bridges the gap between spectral and spatial branches at intermediate depths. Extensive experiments on four benchmark datasets demonstrate that S 2 Fin exhibits good robustness and generalization, and its performance significantly outperforms state-of-the-art methods in few-sample settings. The code is available at https://github.com/HaoLiu-XDU/SSFin .

Large-scale images play a crucial role in geospatial surveying, as they cover an extensively broad view and diverse objects. Due to computational limitations, existing methods rely on generating large-scale images in patches. However, the lack of global guidance in these methods often leads to significant logical errors among different patches. To address this issue, we propose a Global Consistency Diffusion model (Glob-Diffusion) for large-scale image generation. The core idea is to utilize the global consistency of small-scale images to guide the generation of large-scale images. Specifically, we introduce a Hierarchical Distributed Guidance (HDG) module that extracts patch prompts with different semantic hierarchies from small-scale images, distributedly embedding them into the generation of large-scale images to maintain global consistency across various regions. In addition, we further design a Region Guided Adapter (RGA) that dynamically optimizes the guidance strength of patch prompts by comparing differences across generated regions, effectively improving the realism of large-scale images. Our method demonstrates remarkable visual synthesis results across various natural scenes, effectively preserving global consistency in large-scale images, and also significantly enhancing the generation quality of large-scale remote sensing images. Code will be available at https://github.com/kyh433/Glob-Diffusion.

Accurate change detection (CD) in multitemporal multimodal remote sensing images is crucial for numerous applications. However, existing unsupervised CD methods often face challenges in suppressing background noise, preserving fine-grained boundaries, and maintaining spatial coherence of target regions. To overcome these limitations, this study proposes a novel Scribble-Guided Structural Regression Fusion (SG-SRF) framework, which integrates sparse scribble annotations as lightweight priors into a dynamic regression mechanism. Specifically, the framework employs a scribble distance map to refine hypergraph Laplacian matrices, thereby optimizing feature representation for critical targets while suppressing irrelevant backgrounds. Experimental results demonstrate that the proposed method significantly outperforms traditional unsupervised methods in detecting complete and accurate change objects with minimal scribble input. Notably, the scribble guidance offers an efficient and cost-effective solution to the inherent limitations of unsupervised approaches, enabling more precise change detection without extensive labeled datasets. This work aims to bridge the gap between unsupervised adaptability and supervised accuracy, offering significant potential for practical CD applications. The source code will be made publicly available at https://github.com/MissYongjie/SG-SRF.

Radar sounders have become crucial instruments for subsurface exploration in planetary missions, providing insights into the internal structure and composition of celestial bodies. This paper presents a novel automatic framework for the quick-look analysis of radar sounder data acquired in space missions. The methodology serves the dual purpose of supporting both data science analysis and operational planning. By integrating standard and novel processing techniques and evaluation metrics, the framework provides automated assessment of signal quality, early detection of instrument anomalies, and preliminary identification of surface and subsurface features. Initially developed for the Radar for Icy Moon Exploration (RIME) instrument onboard ESAs JUpiter ICy moons Explorer (JUICE) mission, it is designed to be adaptable to other planetary radar sounders, offering a scalable approach to future space exploration missions. The proposed methodology has been experimentally validated using the RIME end-to-end simulator, through a comprehensive set of simulations based on realistic acquisition scenarios representative of different mission phases and target environments.

The increasing number of operating radar sounders (RS) requires fast and reliable automatic data processing methods. In this context, the semantic segmentation of radargrams enables the identification of key subsurface features. While deep learning (DL) is increasingly employed to enhance generalization, few approaches treat radargrams as sequences of correlated columns, despite the natural continuity between adjacent rangelines. Previous sequential methods model this correlation statistically, using techniques such as regression or Markov chains. Moreover, only a few models aim to reduce the number of labeled samples required during training. In this paper, we propose a methodology for the weakly-supervised semantic segmentation of radar sounder data, introducing a novel convolutional recurrent bottleneck to process radargrams. This recurrent unit allows a U-shaped architecture to capture horizontal correlations across the radargram, which is represented as a sequence of columns. Crucially, it also enables the application of newly introduced cycle-consistency-based loss functions, allowing training with just one labeled column per sequence, significantly reducing labeling requirements. We further enhance our architecture with a fast non-local operation to encode the vertical relationships between semantic classes. We validate our approach through extensive experiments on two datasets of radargrams acquired in Antarctica. Results demonstrate the effectiveness of the proposed methodology in weakly-supervised settings and confirm the competitiveness of our architecture even under full supervision.

The unavailability of pixel-level detailed labels is a crucial challenge in the field of remote sensing image analysis. Deep learning models require a large number of labeled samples for an accurate estimation of a large number of trainable parameters. However, in remote sensing applications usually only a few reliable labeled data are available for the learning of a classifier, whereas often many weak/low-resolution unreliable labeled data can be collected from available land-cover maps. Accordingly, weak supervised learning may overcome the problems by using noisy and low-resolution labels in remote sensing. In this paper, we propose a deep adversarial model based on discrete wavelet transform to exploit weak/low-resolution label information for generating refined super-resolved Weak Reference Maps (WRM). Our contribution includes the development of a discrete wavelet transform based generator for enhancing the low-resolution labels to generate a refined high-resolution reference map. We also present an efficient framework for multi-source image fusion that incorporates the refined super-resolved WRM, synthetic aperture radar images and corresponding low-resolution labels. Our findings highlight the effectiveness of the refined super-resolved WRM. Additionally, we investigate the impact of the high-resolution reference maps on segmentation accuracy, which reveals their potential in improving the segmentation performance compared to other reference methods.

Craters are the most typical geologic structures and landforms on the surface of Mars. Martian craters are widely distributed in a variety of morphology with multiple types and exhibit significant differences in scale. Many attempts have been made to automatic identification of Martian craters, yet existing methods do not satisfy the need for large-scale identification. In this article, we first integrate a Martian crater dataset related to the mid- and low-latitude regions, which contains different types and various scales of craters. Then, a dual convolutional neural network (CNN)-Transformer-based cross-modality adaptive feature fusion network (DCT-CMAFFNet) is proposed for accurate identification of the multitype and multiscale craters at large scale on the Martian surface. The proposed network takes full advantage of the rich morphological features contained in Martian imagery and the topographical information reflected by the digital elevation model (DEM) data. It contains two modules: one is the dual CNN-Transformer part, which employs a hybrid architecture to extract the local detailed and global deep features of Martian craters from images and DEM and the other is CMAFF module, which exploits self-attention mechanism to learn the relationship between the images and DEM modalities and weigh each position of the deep feature maps to ensure a comprehensive identification of multitype and multiscale Martian craters. By adaptively fusing the rich information from imagery and DEM, the proposed network identified 3166 new Martian impact craters larger than 1 km, achieving a 14%–24% improvement in accuracy compared to methods using either a single data source or data feature fusion modality.

Analyzing radargrams obtained from radar sounder (RS) instruments is an effective method for studying the subsurface of celestial bodies. However, existing deep learning-based methods for automatic classification of subsurface in RS data require a large amount of labeled training data, posing a significant challenge. Moreover, these methods lack generalization to radargrams from different campaigns due to varying distributions within the same classes. To address these limitations, we propose a novel few-shot pixel-based classification framework for RS data. This framework aims to learn underlying patterns using only a few labeled support samples and adapt quickly to radargrams from different campaigns with minimal labeled information without requiring retraining. Given the scarcity of labeled RS data and the significance of low-level features (such as texture and intensity) in class differentiation over high-level features (such as shapes), we propose simplifying the segmentation task into a classification problem by treating each pixel independently. To preserve spatial information, we introduce spatial input (SI) by integrating neighboring pixels in the depth dimension and incorporating sequence awareness to address misinterpretations of reflections from different subsurfaces. Furthermore, unlike conventional semantic segmentation approaches relying on encoder-decoder structures, our framework eliminates the need for the decoder component. We evaluate the proposed method on different datasets acquired in regions of Antarctica by MCoRDS. Results show the effectiveness and generalization capability of the proposed framework to accurately segment different subsurface targets under very limited data.

Semantic change detection (SCD) involves the simultaneous extraction of changed regions and their corresponding semantic classifications (pre- and post-change) in remote sensing images (RSIs). Despite recent advancements in vision foundation models (VFMs), the fast-segment anything model has demonstrated insufficient performance in SCD. In this article, we propose a novel VFMs architecture for SCD, designated as VFM-ReSCD. This architecture integrates a side adapter (SA) into the VFM-ReSCD to fine-tune the fast segment anything model (FastSAM) network, enabling zero-shot transfer to novel image distributions and tasks. This enhancement facilitates the extraction of spatial features from very high-resolution (VHR) RSIs. Moreover, we introduce a recurrent neural network (RNN) to model semantic correlation and capture feature changes. We evaluated the proposed methodology on two benchmark datasets. Extensive experiments show that our method achieves state-of-the-art (SOTA) performances over existing approaches and outperforms other CNN-based methods on two RSI datasets.

Improving hyperspectral image (HSI) semantic segmentation by exploiting complementary information from supplementary modalities (termed X-modality) is promising but challenging due to significant differences in imaging sensors, image content, and resolution. Existing methods often underutilize the unique spatial–spectral features of HSIs by processing them uniformly with X-modality data. In addition, current cross-modality fusion strategies often suffer from limited intermodal interaction or significantly increased model complexity. To address these limitations, we propose CoMiX, an asymmetric encoder-decoder architecture with deformable convolutions (DCNs) for HSI-X semantic segmentation. CoMiX includes an encoder with two parallel, interacting backbones and a lightweight all-multilayer perceptron (ALL-MLP) decoder. The encoder consists of four stages, each incorporating 2D DCN blocks for the X-modality to accommodate geometric variations and 3D DCN blocks for HSIs to adaptively capture spatial-spectral features. Each stage also incorporates a Cross-Modality Feature enhancement and eXchange (CMFeX) module and a feature fusion module (FFM). CMFeX exploits spatial-spectral correlations across modalities to recalibrate and enhance modality-specific and modality-shared features, while adaptively exchanging complementary information. Its outputs are subsequently fused in the FFM and propagated to the next stage for further learning. Finally, the ALL-MLP decoder aggregates the fused features from all stages to produce the final predictions. Extensive experiments demonstrate that CoMiX achieves state-of-the-art performance and generalizes well to various multimodal datasets. The CoMiX code will be released soon.

The Martian surface landforms are highly related to the safe landing and traversability of Mars rovers. Furthermore, landforms associated with the presence of water/ice, minerals and biosignatures can provide valuable insights for Mars exploration missions, particularly in relation to the selection of landing or sample collection sites. The small number of Martian landform datasets and the scarcity of labelable landform samples over Mars make the precise mapping of Martian landforms a challenging task. In this article, we propose a stepwise deep feature transfer (SDFT) model for the mapping of Martian landforms with a small number of labeled samples. The SDFT model comprises two transfer steps. In the first transfer step, a deep learning model trained on a large public source dataset from Earth is transferred to a medium sized public dataset from Mars. This transfer is conducted through a standard pre-training and fine-tuning procedure utilizing a linear classifier. In the second transfer step, the model is further transferred to a small number of target datasets on Mars through a pre-training and fine-tuning procedure with a cosine distance classifier. The stepwise training technique mitigates the challenges associated with varying datasets and small training samples. The proposed SDFT model has been validated on two self-built sample sets using images from the Mars Reconnaissance Orbiter’s Context Camera (CTX). It has also been employed for landform mapping in two local regions with small samples to evaluate its effectiveness in comparison with existing state-of-the-art methods.

Multimodal remote sensing change detection (MRS-CD) is essential for many geospatial applications. However, it remains challenging due to spectral heterogeneity, temporal misalignment, and inconsistencies between different data sources. Existing MRS-CD methods, particularly those employing graph structures, often suffer from parameter sensitivity, limited generalization across datasets, and difficulty in detecting fine-grained changes. Meanwhile, current weakly supervised approaches typically rely on image-level annotations, which often result in coarse localization, blurred boundaries, and limited transferability across heterogeneous modalities of changes. In this work, we present SHIFT (Scribble-driven Hybrid Iterative Feature Tracker), a weakly supervised MRS-CD approach guided by sparse scribble annotations. SHIFT employs a scribble-driven feature extraction module to extract multimodal features, converting scribble annotations into smooth attention maps with a learnable Gaussian blur module to focus the model on potential change areas. Through iterative refinement, SHIFT enhances detection confidence in change regions. Experiments on 13 public datasets demonstrate the superior performance of the proposed approach compared to a number of state-of-the-art methods. Our approach achieves high accuracy with minimal annotations and robust generalization across diverse multimodal scenarios. The source code will be made publicly available at https://github.com/MissYongjie/SHIFT.

Radar Sounder (RS) instruments are crucial for planetary exploration, as they can penetrate surfaces and reveal subsurface geological features. The next ESA’s EnVision mission, scheduled to be launched in 2031 for the exploration of Venus, will carry the Subsurface Radar Sounder (SRS) to profile the surface crust at low frequencies and increase our understanding of the geological history of Venus. In this context, data simulators play a key role in the prediction and interpretation of instrument-specific results, modelling different geoelectrical and morphological features in the analysed scenario. In this paper, we present a study of the SRS’s ability to detect the subsurface structures of lava flows which are one of the targets of interest. This is accomplished by examining morphological features such as their thickness and their geoelectrical properties through simulations. The results show that SRS has favourable conditions to achieve the scientific goals in detecting lava flow subsurface structures.

Highly accurate estimates of canopy height (CH) and above ground biomass (AGB) are key parameters for forest disturbance analysis, resource monitoring, and carbon flux analyses. In this work we present a deep learning-based approach for mapping CH and AGB on country-scales from single-baseline, single-polarization, single-pass TanDEM-X InSAR data. The proposed approach consists in a convolutional neural network (CNN), trained and validated on the five test-sites covered by the 2016 AfriSAR campaign. The resulting performance is in line or better than those of current state-of-the-art approaches. The framework is subsequently deployed on a large-scale to map the entire country of Gabon (in West Central Africa), showcasing the flexibility, scalability, and accuracy of our proposed approach for forest parameter estimation.

Radar sounders (RS) are electromagnetic sensors employed for subsurface investigation on Earth and other celestial bodies. The data acquired by RS are represented as radargrams and they are traditionally analyzed through visual interpretation or manual feature engineering to detect subsurface layers and relevant features. The emergence of deep learning has prompted exploration into deep automated radargram analysis, frequently treating them as conventional images. Here we present a novel method for the semantic segmentation of radar-grams inspired to video object segmentation (VOS). It consists in an initial phase of self-supervised VOS-based learning of features from sequences of radargram patches, and a second phase of label propagation from an initial reference patch provided by experts. The effectiveness of the proposed method is confirmed by validation on a dataset of radargrams generated by the airborne radar sounder MCoRDS1 in Antarctica.

In this study, we explored the use of long time series of Sentinel-1 SAR images for the detection of bark beetle outbreaks in temperate forests. Bark beetle attacks induce a gradual deterioration in the health of trees, ultimately leading to their death. Accordingly, time series of remote sensing data are crucial to detecting them. In this study, time series of Sentinel-1 data were collected in reference areas attacked by bark beetle in the period 2019-2022 and then the backscatter values were extracted and analyzed. A method for the automatic detection of bark beetle attacks has been developed and tested on the same polygons. The preliminary results showed that the distribution of the Sentinel-1 backscatter values in VV and VH polarizations inside the attacked polygons before and after that attack is significantly different. The detection method achieved a detection accuracy above 65% for all the dates considered.

The temporal consistency of yearly land-cover maps is of great importance to model the evolution and change of the land cover over the years. In this paper, we focus the attention on a novel approach to classification of yearly satellite image time series (SITS) that combines deep learning with Bayesian modelling, using Hidden Markov Models (HMMs) integrated with Transformer Encoder (TE) based DNNs. The proposed approach aims to capture both i) intricate temporal correlations in yearly SITS and ii) specific patterns in multiyear crop type sequences. It leverages the cascade classification of an HMM layer built on top of the TE, discerning consistent yearly crop-type sequences. Validation on a multiyear crop type classification dataset spanning 47 crop types and six years of Sentinel-2 acquisitions demonstrates the importance of modelling temporal consistency in the predicted labels. HMMs enhance the overall performance and F1 scores, emphasising the effectiveness of the proposed approach.

Addressing crop-mapping problems using high-resolution hyperspectral images requires innovative solutions that integrate advanced image processing and machine learning. In this paper, we propose an approach designed to address the main challenges of crop mapping using hyperspectral images, including the analysis of high-dimensional feature spaces, the limited number of training samples and the complex spectral-spatial relationships. The proposed architecture exploits Contractive-Expansive-Contractive (CEC) connections, which extract features capable of capturing both global and local patterns, while enhance discriminative features and suppress noise and irrelevant information. The CEC network is used to integrate multiscale features and learn the optimal decision boundaries for segmentation. The proposed architecture is evaluated on a Prisma hyperspectral dataset and compared against other methodologies suited for crop-type mapping. The experimental results demonstrate that the proposed approach achieves higher overall and class-wise segmentation accuracy.

Very high-frequency (VHF) radar sounding from an orbital platform is a promising mission concept for subsurface observations of the Earth’s polar caps for mapping the ice sheets and ice shelves from the surface to the base. Previous feasibility studies have analyzed the detectability of the basal interface, the internal layers and the subglacial lakes in simulated VHF radargrams. However, there are few studies on the detection of near-surface melt-water and the ice-ocean interface, which are important for predicting the stability of the polar ice. This paper both presents a novel semi-supervised basal ice-water detection algorithm and exploits a state-of-the-art dielectric inversion technique to detect surface melt for the analysis of simulated VHF radargrams. Results show that about 90% of the ice-ocean interfaces are correctly detected with the proposed detection algorithm and that the dielectric inversion revealed a pool of melt-water in the Antarctic peninsula.

The development of satellite and airborne sensor technologies has resulted in a wealth of multisource remote sensing images, which have great potential for precise analysis and monitoring of the Earth’s surface. Despite recent advancements in change detection (CD) on multitemporal remote sensing images, there are still many challenges in effectively integrating multimodal remote sensing data in the CD task. To address these challenges, this paper proposes a novel unsupervised Transformer-enhanced Encoder-Decoder CD (namely ETD-CD) framework for heterogeneous remote sensing images. The proposed framework focuses on fusion of multimodal features for achieving high-precision CD results without relying on high-quality manual labels. Experimental results demonstrate the superiority of the proposed ETD-CD network for multimodal remote sensing image CD.

Pits are depressions in the ground that occur due to the collapse of the surface layer. The characterization from orbit of their internal structure using optical images is challenging due to uncontrolled illumination geometry. In this paper, we propose a methodology for the characterization of pits’ interiors by exploiting Synthetic Aperture Radar (SAR) images. The methodology analyzes the amplitude and range of the radar echoes originating from the pit’s interior for determining its geometric characteristics through data inversion. The experimental results demonstrate that a set of bright reflections in the radar image can be attributed to the response of pits’ vertical walls and floor. By applying the proposed methodology and interpreting the radar reflections, we are able to derive a geometric characterization (e.g., depth) of a given pit. The retrieved geometric parameters from SAR data of a terrestrial pit denoted asWell of Barhoutalign well with the ground truth. The findings of this study have implications for both Earth observation and planetary exploration.

In planetary explorations, missions that employ flybys operate in strongly variable environments. The rapid change in altitude of a flyby orbit increases the difficulties in determining the optimal instrument parameters to properly study the investigated celestial body. Typically, the selection of the parameters is guided by multiple criteria according to the mission’s scientific objectives and the available resources. In this letter, we propose a novel automatic strategy to find the optimal instrument configuration exploiting a multiobjective optimization. The problem is addressed as the minimization of an objective function, which evaluates the acquisition properties. Thus, the solutions represent the best trade-offs between system performance and resource consumption. We tested our approach on a monostatic radar sounder (RS) capable of penetrating and analyzing the celestial bodies’ subsurface. In particular, we considered the Radar for Icy Moon Exploration (RIME) on board the JUpiter Icy moons Explorer (JUICE) that will study the Jovian moons: Europa, Ganymede, and Callisto. Several planning simulations have been conducted, considering numerous flybys on the three moons while adopting signal-to-noise ratio (SNR), data rate, and orbit uncertainty as objectives. Results confirmed the effectiveness of the proposed approach, reaching a surface SNR of 60 dB with a data rate of 9 Mb/s for the 7E1 flyby on Europa.

In recent years, deep learning methods, in particular Convolutional Neural Networks (CNNs), have been increasingly used in Change Detection (CD). However, most CNN-based CD methods are primarily designed for analyzing only a single pair of images due to the challenge of collecting and constructing ground reference data during the system-training phase. Consequently, existing CD methods, particularly those focused on detecting multi-annual changes, exhibit limited capability in extracting comprehensive spatio-temporal information. To address this limitation, we propose a novel weakly supervised deep learning-based technique for CD exploiting a 3D CNN architecture to extract spatio-temporal information. Our technique incorporates a fine-tuning stage to effectively capture temporal patterns from a yearly Satellite Image Time Series (SITS) by using different 3D convolutional layers. It also exploits a multi-feature hyper-temporal Change Vector Analysis (CVA) for multi-annual change identification. The proposed approach is tested on a four year dataset in Amazonia and gained the highest yearly CD accuracy of 88.59%, 97.27% and 87.87% for 2017, 2018 and 2019, respectively.

Spectral-spatial classification of hyperspectral images (HSIs) has been extensively studied. Although the importance of spatial information for classification of HSIs is widely proven in the literature, the definition of effective techniques for the extraction of spatial information is still a challenging and open research issue. In this letter, a semantic edge-aware structure preserving image filtering technique is presented to accurately model spatial information in HSI classification. The experimental results on the three real HSI data sets show the superiority of our model, which provides at least 2% higher classification accuracy than the best among the numerous literature models considered.

Shallow aquifers are the primary water source to mitigate rising hydroclimatic fluctuations in arid areas, notably in North Africa and the Arabian Peninsula. The occurrence and dynamics of these expansive water bodies remain poorly characterized due to the reliance on sporadic monitoring wells. To address this deficiency, several studies are exploring the potential of low Earth orbit sounding radars as a large-scale mapping tool that can provide unique insights into the delineation and dynamics of these aquifers. Herein, we analyze the detectability of shallow aquifers (<10 m deep) using the radiometric analysis of surface reflections from a 45-MHz orbital sounder with an 8-MHz bandwidth. We use the ray tracing method to simulate the radar return from two realistic geoelectrical and topographic models of shallow aquifers in North African Sahara desert for omnidirectional and distributed array configurations. Our results suggest that the dielectric change induced by shallow aquifers that are up to 10-m deep can increase the 45-MHz radar surface return of the desiccated desert surface by 5 dB in areas with very low surface roughness of rms height <0.35 m. These preliminary results suggest a constrained potential for a monostatic VHF reflectometry to probe large sedimentary basins, which a distributed architecture can improve.

Deep learning (DL) approaches are widely used to improve change detection (CD). In many application domains, unsupervised DL CD methods are preferred since gathering multitemporal labeled samples is challenging. Many unsupervised CD methods use pre-trained DL models to extract multiscale features. This does not allow for preserving the spatial context information and the object structure in the multiscale hidden-layer features, thus obtaining poor performance in modeling multiresolution changes. In this article, we propose two hierarchical loss functions to train multiscale hidden-layer features and preserve their spatial context information. The multiscale hidden-layer feature maps extracted from the model are used in an unsupervised multiscale CD method. We present two possible hierarchical loss functions. The first one considers all the model layers during the training by comparing the mirrored couples of encoder–decoder hidden-layer features, while the second one aims to preserve the geometrical details using a multiresolution-based loss function. After training, the CD method uses a feature selection (FS) based on structure-similarity-index (SSIM) to keep only the most informative hidden-layer feature maps. We tested the proposed method on bi-temporal multispectral images acquired by Landsat-8 representing a burned area and Sentinel-2 images representing a deforested area.

The comprehension of pit craters on Venus’ surface is still limited both in terms of genetic process and geometric characteristics. Pit craters are valuable features for planetary scientists and geologists as they offer a window into a planetary body’s history, geological processes, and environmental conditions. In this context, morphometry is a valuable tool for unraveling the geological history of these structures and for comparative investigations with analogous features occurring on other celestial bodies. In this study, we present a methodology utilizing orbital synthetic aperture radar (SAR) data to estimate the characteristics of Venusian pits. Our findings demonstrate that it is feasible to estimate the depth of a given pit crater by analyzing reflections from the pit’s wall facing the radar illumination direction. This method offers an additional estimate of the pit’s depth, complementing literature techniques based on radar shadows, which may not always be available depending on the acquisition geometry. Depending on the configuration of the pit’s interior structure, the depth estimation derived from the pit wall’s reflections is potentially less prone to geometric distortions than the one obtained from the radar shadow. The morphometric analysis of Venusian pits, based on the radar-derived parameters from Magellan SAR data, reveals that these pits exhibit very large collapse volumes and remarkable morphological similarities with lunar pit craters. This discovery has implications for comparative planetology, shedding light on potential shared mechanisms and processes governing the formation and evolution of such features.

Unsupervised change detection using contrastive learning has significantly improved the performance of literature techniques. However, at present it only focuses on the bi-temporal change detection scenario. Previous state-of-the-art models for image time-series change detection have traditionally depended on features obtained either through clustering learning or by training models from scratch using pseudo labels tailored to each scene. However, these approaches fail to either exploit the spatial-temporal information of image time-series or generalize to unseen scenarios. In this work, we propose a two-stage approach to unsupervised change detection in satellite image time-series using contrastive learning with feature tracking. By deriving pseudo labels from pre-trained models and using feature tracking to propagate them within the image time-series, we improve the consistency of our pseudo labels and address the challenges of seasonal changes in long-term remote sensing image time-series. We adopt the self-training algorithm with ConvLSTM on the obtained pseudo labels, where we first use supervised contrastive loss and contrastive random walks to further improve the feature correspondence in space-time. Then a fully connected layer is fine-tuned on the pre-trained multi-temporal features for generating the final change maps. Through comprehensive experiments on two datasets, we demonstrate consistent improvements in accuracy on fitting and inference scenarios.

High-resolution digital terrain models (DTMs) are critical for supporting planetary exploration missions and advancing scientific research. Recently, deep learning (DL) techniques have been applied to reconstruct high-resolution DTMs from single-view orbiter optical images, particularly for the Moon. However, DL-based methods face challenges in retrieving high-quality multiscale topographic features, especially in regions with irregular terrains or significant relief. Additionally, their generalization capability across diverse datasets is rarely evaluated. In this article, we propose an efficient DL-based single-view method with a coarse-resolution DTM as a constraint for high-quality lunar DTM reconstruction, named ELunarDTMNet. This approach introduces a hierarchical transformer-based backbone with a residual-connected mechanism, specifically designed to capture and integrate multiscale features from single-view lunar images, thereby enhancing prediction accuracy. Meanwhile, given the diverse and complex surface relief, new elevation normalization strategies are proposed to preserve terrain feature contrast while accommodating different elevation distributions. Our method performs well on diverse lunar landscapes with various topographic features and elevation changes. It outperforms the existing DL-based methods in accuracy and detail, effectively addressing their encountered challenges. Moreover, the proposed method achieves effective resolutions similar to those of the shape-from-shading (SFS) technique for subtle-scale terrain retrieval, but with enhanced elevation accuracy, illumination robustness, and approximately$850\times $faster processing speed. Trained with the lunar reconnaissance orbiter (LRO) narrow angle camera (NAC) images, our model shows superior performance on other high-resolution lunar orbiter images, such as Chang’E-2 imagery.

The mechanism of connecting multimodal signals through self-attention operation is a key factor in the success of multimodal Transformer networks in remote sensing data fusion tasks. However, traditional approaches assume access to all modalities during both training and inference, which can lead to severe degradation when dealing with modal-incomplete inputs in downstream applications. To address this limitation, we propose a novel approach to incomplete multimodal learning in the context of remote sensing data fusion and the multimodal Transformer. This approach can be used in both supervised and self-supervised pre-training paradigms. It leverages the additional learned fusion tokens in combination with modality attention and masked self-attention mechanisms to collect multimodal signals in a multimodal Transformer. The proposed approach employs reconstruction and contrastive loss to facilitate fusion in pre-training, while allowing for random modality combinations as inputs in network training. Experimental results show that the proposed method delivers state-of-the-art performance on two multimodal datasets for tasks such as building instance / semantic segmentation and land-cover mapping when dealing with incomplete inputs during inference.

Radar sounders (RSs) are utilized for the analysis of subsurface of Earth and other planets. Data acquired from RS can be processed to obtain radargrams, which are 2-D arrays containing the backscattered echo power received by the radar after sending pulses toward the surface. The study of radargrams offers crucial insights for the geological interpretation of the history of planets and for the monitoring of ice layers in glacial regions. Deep learning (DL) has emerged as a powerful tool for the automatic feature extraction and analysis of radargrams; yet, they are still treated as conventional images. We propose a novel methodology for the semantic segmentation of RS data based on a two-step approach. The rationale of this methodology is exploiting the spatial horizontal correlation that exists among radargram features, which is an important property that distinguishes these data from standard images. In the first step, an encoder is trained in an unsupervised way, exploiting random walks to learn meaningful representations of sequential features within radargrams. In the second step, few reference labeled samples allows the model to propagate the labels to the full radargram. We also introduce a metric to quantify the degree of horizontal correlation among features, and we use it to find the grounding zone in coastal radargrams of polar areas. We test our methodology on two datasets obtained by the multichannel coherent radar depth sounder (MCoRDS) RS and a dataset from the orbital RS shallow radar (SHARAD) and we discuss the very promising results.

Convolutional neural networks (CNNs) have become a popular and powerful tool for polarimetric synthetic aperture radar (PolSAR) image classification. The success of the CNN model is dependent on the features that the networks extract from the polarimetric channels during the learning phase. To extract better discriminative features, we propose a novel two-step method. In the first step, by exploiting singular value decomposition (SVD), a multiscale SVD profile (MSVDP) is constructed that models spatial information of each pixel of the PolSAR image in multiple scales. In the second step, a lightweight and shallow dual-branch CNN is proposed to take the original PolSAR image and the constructed MSVDP as inputs for extracting more discriminative features during the learning of the CNN model. The effectiveness of the proposed model is validated using three real PolSAR datasets. Our proposed technique provides accurate and satisfactory results irrespective of the considered polarimetric feature sets and power descriptors. Source code for the dual-branch CNN is available athttps://github.com/ND-PatternHunter/DB-SVD-CNN

Semantic Change Detection (SCD) refers to the task of simultaneously extracting the changed areas and the semantic categories (before and after the changes) in Remote Sensing Images (RSIs). This is more meaningful than Binary Change Detection (BCD) since it enables detailed change analysis in the observed areas. Previous works established triple-branch Convolutional Neural Network (CNN) architectures as the paradigm for SCD. However, it remains challenging to exploit semantic information with a limited amount of change samples. In this work, we investigate to jointly consider the spatio-temporal dependencies to improve the accuracy of SCD. First, we propose a Semantic Change Transformer (SCanFormer) to explicitly model the ’from-to’ semantic transitions between the bi-temporal RSIs. Then, we introduce a semantic learning scheme to leverage the spatio-temporal constraints, which are coherent to the SCD task, to guide the learning of semantic changes. The resulting network (SCanNet) significantly outperforms the baseline method in terms of both detection of critical semantic changes and semantic consistency in the obtained bi-temporal results. It achieves the SOTA accuracy on two benchmark datasets for the SCD.

Vision Foundation Models (VFMs) such as the Segment Anything Model (SAM) allow zero-shot or interactive segmentation of visual contents, thus they are quickly applied in a variety of visual scenes. However, their direct use in many Remote Sensing (RS) applications is often unsatisfactory due to the special imaging properties of RS images. In this work, we aim to utilize the strong visual recognition capabilities of VFMs to improve change detection (CD) in very high-resolution (VHR) remote sensing images (RSIs). We employ the visual encoder of FastSAM, a variant of the SAM, to extract visual representations in RS scenes. To adapt FastSAM to focus on some specific ground objects in RS scenes, we propose a convolutional adaptor to aggregate the task-oriented change information. Moreover, to utilize the semantic representations that are inherent to SAM features, we introduce a task-agnostic semantic learning branch to model the semantic latent in bi-temporal RSIs. The resulting method, SAM-CD, obtains superior accuracy compared to the SOTA fully-supervised CD methods and exhibits a sample-efficient learning ability that is comparable to semi-supervised CD methods. To the best of our knowledge, this is the first work that adapts VFMs to CD in VHR RS images.

Radar sounder (RS) profiles are essential for imaging the subsurface of planetary bodies and the Earth as they provide valuable geological insights. However, the limited availability of high-resolution radargrams poses challenges. This article proposes a novel method based on generative models to super-resolve radargrams. Our approach addresses the ill-posed and ill-conditioned nature of the super-resolution problem by training a neural network to learn the correlation between radargrams at different scales. The network learns a proxy for the mapping function between ambiguous low-resolution radargrams and more detailed high-resolution ones, considering the data’s geological and statistical properties. The mapping function enables the super-resolution of previously unseen low-resolution radargrams acquired in comparable conditions to those in the training and imaging similar underlying geology. To achieve this, we adopt a cycle generative adversarial network (CycleGAN), explicitly designed to match properties between low- and high-resolution radargrams, accounting for variations in dimensions and radiometric properties. Furthermore, we enhance the network performance by incorporating skip connections, a ResNet module, and attention mechanisms. The proposed method is validated using MCoRDS3 radargrams acquired in Greenland and Antarctica as high-resolution data. As low-resolution data, we used simulated radargrams representing what is expected by an Earth-orbiting low-resolution RS to have a controlled experiment. The results are evaluated qualitatively and quantitatively, focusing on the areas with reflections with complex shapes that may generate artifacts and unrealistic geological features.

The integration of spatial and spectral information is beneficial to the improvement of change detection (CD) performance. However, existing methods cannot efficiently suppress the influences of spatial and spectral differences (SDs) in unchanged areas. To address these issues, in this article, we propose a content-guided spatial–spectral integration network (CSI-Net) for the fusion of global spatial details and SD information. Specifically, the proposed CSI-Net is composed of a spatial reasoning (SR) module, an SD module, and a content-guided integration (CGI) module. In the SR module, the spatial information is learned by cascaded graph convolution (GC) blocks for global modeling. The SD module is responsible for the extraction of spectral features, by calculating the means and variances of features to reduce the impact of SDs in unchanged regions. In addition, in order to integrate the spatial–spectral features efficiently, we design a CGI module to further take advantage of their complementary information. In this module, high-level content information is introduced as a guide for proper interaction. Due to the efficient spatial–spectral fusion, the proposed CSI-Net can learn the changed features better while achieving suppression of SDs. Experimental results on LEVIR-CD, WHU-CD, and CLCD datasets demonstrate that the proposed CSI-Net produces better performance compared to state-of-the-art methods, and is applicable to different scenarios. The code of CSI-Net is available athttps://github.com/RSMagneto/CSI-Net.

High-resolution (HR) satellite image time series (SITS) are a valuable data source for analyzing land cover change (LCC) due to their large amount of spatial, spectral, and temporal information. However, most existing LCC detection methods focus on binary change detection (CD) within a single year and fail to provide detailed information about the specific type of change. In this study, we propose a multiannual CD approach that identifies changes occurring between consecutive years and provides information about the type of LC transition. The proposed approach exploits multiannual and multispectral SITS to generate a hypertemporal feature space (FS). This FS is analyzed to create a set of CD maps that indicate the time, probability, and type of change. To measure the similarity between pixel time series, we use dynamic time warping (DTW) in the space of hypertemporal features. A hierarchical clustering technique is exploited to develop a set of class prototypes (CPs) that represent the characteristics of different LC classes. The CPs are then used to identify the most probable LC transition for each changed pixel. Two test areas were selected to evaluate the effectiveness of the proposed approach. The first one is located in Amazon and spans the years 2015 to 2019; and the second one is located in Sahel-Africa and covers the years 2015 and 2016, using multiannual Landsat 7 and 8 SITS. The results demonstrate that the proposed approach is effective in detecting multiannual changes and in identifying the LC transitions.

Convolutional neural networks (CNNs) have performed notably in change detection (CD) tasks due to their superior learning and automatic feature extraction capabilities. However, they suffer from the limited receptive field and the weak modeling of long-range dependencies. Vision transformers (ViTs) excel in modeling long-range contexts and have been recently introduced in CD. Some works have combined CNN and transformers to obtain local-global information. However, these works do not fully consider the guidance and interactions from both local features (LFs) and global features (GFs). Most importantly, most of them involve a very large number of parameters and computational costs. To address these issues, in this article, we propose a lightweight CD network that mixes features across CNN and transformer (MixCDNet). We use EfficientNet as the backbone and design a novel mixing features block (MFB). First, we employ hierarchical feature extraction blocks, where local feature blocks (LFBs) and global feature blocks (GFBs) are utilized for extracting information at different spatial resolutions. Second, we propose to exploit bidirectional interactions across LFBs and GFBs branches to provide complementary clues while capturing LFs and GFs. Moreover, a skip-connection and fusion separable self-attention layer (SFSSL) is designed to obtain GFs with low complexity. Comprehensive experiments are conducted on three high-resolution remote sensing (HRRS) images CD datasets: LEVIR-CD, WHU-CD, and CDD. The results show the effectiveness of the proposed MixCDNet in improving the performance of existing CD methods with fewer parameters (0.32 M) and lower computation costs (1.59G FLOPs).

Hyperspectral image (HSI) classification has recently reached its performance bottleneck. Multimodal data fusion is emerging as a promising approach to overcome this bottleneck by providing rich complementary information from the supplementary modality (X-modality). However, achieving comprehensive cross-modal interaction and fusion that can be generalized across different sensing modalities is challenging due to the disparity in imaging sensors, resolution, and content of different modalities. In this study, we propose a local-to-global cross-modal attention-aware fusion (LoGoCAF) framework for HSI-X segmentation. LoGoCAF adopts a two-branch semantic segmentation architecture to learn information from HSI and X modalities. The pipeline of LoGoCAF consists of a local-to-global encoder and a lightweight all multilayer perceptron (ALL-MLP) decoder. In the encoder, convolutions are used to encode local and high-resolution fine details in shallow layers, while transformers are used to integrate global and low-resolution coarse features in deeper layers. The ALL-MLP decoder aggregates information from the encoder for feature fusion and prediction. In particular, two cross-modality modules, the feature enhancement module (FEM) and the feature interaction and fusion module (FIFM), are introduced in each encoder stage. The FEM is used to enhance complementary information by combining information from the other modality across direction-aware, position-sensitive, and channel-wise dimensions. With the enhanced features, the FIFM is designed to promote cross-modality information interaction and fusion for the final semantic prediction. Extensive experiments demonstrate that our LoGoCAF achieves superior performance and generalizes well on various multimodal datasets. Code is available athttps://github.com/xumzhang.

In this article, we propose a new pan-sharpening method that disentangles low spatial resolution multispectral (LRMS) and panchromatic (PAN) images in terms of sensor-specific features and common features. These features are obtained by defining mutual information (MI)-based transformers designed to achieve disentangled learning. In the proposed method, LRMS and PAN images are cross-reconstructed by cross-coupled transformers to facilitate the disentanglement of the common features and sensor-specific features. To ensure compatibility among the disentangled features, self-reconstructions of LRMS and PAN images are imposed on them, and source images are reconstructed by self-coupled transformers. In addition to the reconstruction-guided disentangled learning, we maximize the MI between the common features of LRMS and PAN images to improve the correlation of the common features from different images. We also minimize the MI between the common features and sensor-specific features from the same image to reduce the redundancy among them. Through the reconstruction and disentangled representation of source images, sensor-specific features and common features can be decomposed efficiently. Finally, all disentangled features are integrated by a fusion transformer to generate the high spatial resolution multispectral (HRMS) image. Experiments on different datasets demonstrate that the proposed method produces competitive fusion results. The code is available athttps://github.com/RSMagneto/DRFormer.

Change detection (CD) in high-resolution remote sensing has received large attention due to its wide range of applications. Many methods have been proposed in the literature and achieved excellent performance. However, they are often fully supervised, thus requiring abundant pixel-level labeled samples, which is time-consuming and labor-intensive. Especially compared to the common single-temporal interpretation, labeling bi-temporal images is often more complicated. Therefore, this study combines weakly supervised learning (WSL) to reduce label acquisition costs. But changed regions are small, fragmented, and similar to the background, which increase the gap between weakly supervised and fully supervised tasks. To address these difficulties, we explore self-supervised methods to construct a WSL framework based on image-level labels for general CD, termed WSLCD in this paper. First, we design a double-branch siamese network to derive embeddings and initial class attention maps (CAMs), which inputs the original image pair and the spatially transformed image pair. Second, mutual learning and equivariant regularization (MLER) is enforced on CAMs from different views, which implements consistency constraints in confusion regions and makes CAMs learn from each other based on saliency regions. Furthermore, prototype-based contrastive learning (PCL) is designed such that unreliable pixels can learn from prototypes computed from reliable pixels. PCL includes intra-view contrast and cross-view contrast depending on whether the prototypes and class embeddings are from the same view. With the above strategies, we narrow the gap between image-level weakly supervised CD and fully supervised CD. Experiments are conducted on three CD datasets, including CLCD, DSIFN and GCD. Our method achieves state-of-the-art performance on pseudo label generation and CD. The code is available at https://github.com/mfzhao1998/WSLCD.

Landform classification and mapping of the Martian surface using Mars orbiter images can provide an important reference for landing site selection and rovers’ traversability evaluation in Mars exploration. Moreover, specific Martian landforms are closely associated with the evidences of water-related activities and Martian life, thus have crucial research importance. This article proposes a novel superpixel-guided multiview feature fusion network (MarsMapNet) for efficient mapping of the Martian landforms. In particular, the proposed MarsMapNet first generates the superpixel-level segments from Mars orbiter images by considering local morphological homogeneity of landforms. Then, a multiview feature extraction and fusion (MVF) network is developed, where abstract convolutional features are extracted based on scene-level patches, and multitextures are extracted based on local landform from shallow-to-deep feature learning. After the network being trained on scene-level samples and guided by the superpixel segmentation, Martian landforms can be correctly classified in an efficient way, whose mapping time cost sharply decreased when compared to the reference methods. The proposed MarsMapNet has been validated on three real landing sites from several Mars missions (i.e., the Jezero Crater, the Southern Utopia Planitia, and the Oxia Planum) by using the Mars Reconnaissance Orbiter’s Context Camera (CTX) images. Qualitative and quantitative analyses on the obtained experimental results confirm the effectiveness and efficiency of the proposed MarsMapNet when compared with the state-of-the-art (SOTA) methods, demonstrating its potential for supporting a Martian global landform mapping in the future.

With the increasing number of high resolution (HR) images captured by various platforms, integrating spectral and spatial properties of data across different HR image types, such as multispectral (MS), hyperspectral (HS), and multitemporal (MT) images, remains a challenging task for object classification. This paper proposes a novel hybrid framework named Hybrid FusionNet (HFN) that jointly exploits 2D-3D Convolutional Neural Networks (CNNs) and a Transformer encoder to address a complex classification problem. By incorporating 2D and 3D convolutional layers, the proposed HFN generates rich multi-dimensional hybrid features, including spectral, spatial, and temporal features. These features are then fed into a Transformer encoder to learn global saliency and discriminative information, enabling the identification of spatially irregular and spectrally similar objects. The hybrid architecture efficiently captures local intricate spectral-spatial-temporal contextual features through convolutional layers. Then it learns global long-range dependencies and the spectral dimension through the Transformer encoder, thus effectively reducing spectral-spatial mutations, distortions, and variations of ground objects. Experimental results from an HR-MS dataset, an HR-HS dataset, and an HR-MT dataset covering complex urban scenarios confirm the effectiveness of the proposed approach compared to the main state-of-the-art methods. Notably, the proposed HFN can achieve satisfactory classification performance even with limited training samples. The source code will be made available at https://github.com/MissYongjie/Hybrid-FusionNet.

Fusion of a panchromatic (PAN) image and corresponding multispectral (MS) image is also known as pansharpening, which aims to combine abundant spatial details of PAN and spectral information of MS images. Due to the absence of high-resolution MS images, available deep-learning-based methods usually follow the paradigm of training at reduced resolution and testing at both reduced and full resolution. When taking original MS and PAN images as inputs, they always obtain sub-optimal results due to the scale variation. In this paper, we propose to explore the self-supervised representation for pansharpening by designing a cross-predictive diffusion model, named CrossDiff. It has two-stage training. In the first stage, we introduce a cross-predictive pretext task to pre-train the UNet structure based on conditional Denoising Diffusion Probabilistic Model (DDPM). While in the second stage, the encoders of the UNets are frozen to directly extract spatial and spectral features from PAN and MS images, and only the fusion head is trained to adapt for pansharpening task. Extensive experiments show the effectiveness and superiority of the proposed model compared with state-of-the-art supervised and unsupervised methods. Besides, the cross-sensor experiments also verify the generalization ability of proposed self-supervised representation learners for other satellite datasets. Code is available at https://github.com/codgodtao/CrossDiff.

Large-scale and up-to-date canopy height model (CHM) estimates are key to forest resources assessment and disturbance analysis. In this work we present an investigation of the potential of Deep Learning (DL) for the regression of forest height from TanDEM-X bistatic InSAR data. We propose a novel fully convolutional neural network (CNN) framework, trained and tested on four tropical sites in Gabon, Africa, together with a series of experiments for assessing the impact of different input features with specific focus on bistatic InSAR. The obtained results are extremely promising and already in line with state-of-the-art methods based on theoretical modelling, with the remarkable advantage of requiring only one single TanDEM-X acquisition at inference time.

Circular sinkholes (i.e., pits) are depressions in the ground caused by the collapse of the surface layer. These type of features are commonly found on Earth and on other celestial bodies such as the Moon and Mars. Sinkholes may provide access to an underground cave system but it is very difficult to understand their accessibility and internal shape from optical images. In this paper, we propose a methodology for characterizing circular sinkholes by exploiting X-Band Very High Resolution (VHR) Spaceborne Synthetic Aperture Radar (SAR) images. The analysis of our experimental results show that a set of bright reflections in the radar image can be attributed to the response of the circular pit vertical walls and interior. Accordingly, it is possible to characterize the internal morphology of the pit. The results of this work have implications for planetary exploration as the proposed methodology can be applied for characterizing planetary pits.

In this work, we address the challenge of open-world semantic segmentation for remote sensing (RS) images, which involves segmenting arbitrary objects in images using open RS data. Previous efforts in open-world segmentation mostly focus on Internet-scale paired image-text data with rich vocabulary of concepts. However, these works cannot be directly transferred to RS domain due to the lack of large-scale RS data-text pairs and the corresponding annotations. To overcome this limitation, we propose using text descriptions and annotations from OpenStreetMap as a source of supervision while using images from satellite images. We utilize a conditional Unet model to predict segmentation masks given a text description, and leverage the rich information contained in a pretrained CLIP model to align the images and the corresponding text embeddings using a contrastive loss. Our experimental results demonstrate the potential of open-world segmentation on open RS data.

Analyzing radar sounder (RS) profiles allows the retrieval of critical information on subsurface geology. However, radar-grams suffer from several noise contributions, adversely affecting the data quality and reliability. In the remote sensing literature, there are no methods for denoising radargrams, and those for optical data denoising and SAR and GPR data de-speckling are based on assumptions that are not valid in the RS domain. This paper analyses the statistical distributions of the noisy contributions in radargrams at different levels of processing. It proposes a novel method to denoise complex raw radargrams using a generative network (diffusion probabilistic model) that learns the noise statistical properties. The model is iteratively trained to learn the information loss as a function of the noise level increment in the data. By reversing the process, the network estimates the noise statistical properties and denoises unseen radargrams. The method has been successfully validated on the raw Experiment Data Record (EDR) radargrams of Mars that the Shallow Radar (SHARAD) acquired.

A great effort has been put on developing technologies that can process High Resolution (HR) satellite datasets to properly monitor the environmental changes and produce long term Change Detection (CD) maps. However, there is still a need to design CD approaches that process Satellite Image Time Series (SITS) with high spatial, spectral, and temporal resolution and describe changes that have occurred between the consecutive years. Here, a CD processing chain is proposed that: i) extracts several relevant features of the spectral trends of different sets of LC changes, ii) produces a regular and dense feature time series, iii) analyzes differences between the consecutive years by using a Multi-feature Hyper-temporal Change Vector Analysis (MHCVA) technique, and iv) detects the year and the probability of changes at pixel level. The effectiveness of the proposed approach is tested on a multi-annual Landsat 7 and 8 images of an area located in Amazon.

Finding pixel-level information is a fundamental problem in computer vision and in remote sensing image analysis. Usually a classifier based on a deep learning model requires a large number of labeled samples for an accurate estimation of a large number of trainable parameters. However in remote sensing applications usually only a few reliable labeled data are available for the learning of a classifier, whereas often many weak/low-resolution unreliable labeled data can be collected from absolute land cover maps. Accordingly, weak supervised learning may overcome the problems by utilising noisy and low-resolution labels in remote sensing. In this paper, we propose a deep adversarial model based on discrete wavelet transform to exploit weak/low-resolution label information for generating refined feature maps. Our method mainly includes a feature learning generative network based on wavelet features and a multiscale modeling capability.

Semantic change detection (SCD) is a branch of change detection (CD) that provides detailed land-cover/land-use (LCLU) change information. It presents not only the changed information but also the bi-temporal LCLU semantic maps (in the changed areas). Studies have recently highlighted [1] that SCD can be addressed through a triple-branch Convolutional Neural Network (CNN), which contains two multi-temporal branches and a change detection branch. However, in this architecture, the two temporal branches learn insufficient LCLU transition information. In this paper, we present a novel architecture that combines CNN and RNN for the SCD of remote sensing images. It employs a Siamese CNN to learn semantic information from two temporal images, followed by a Bidirectional Recurrent Neural Network (Bi-directional RNN) to learn temporal dependencies of the LCLU classes. The resulting CNN-RNN architecture can model better the LCLU transitions, thus enhancing the semantic representation of the bi-temporal features. Experimental results on a benchmark dataset show that the proposed method obtains significant accuracy improvements over the existing approaches. It also shows advantages in recognizing the minority LCLU changes.

Active learning is one of the popular approaches that can mitigate some of the drawbacks of supervised classification. Although sparse representation classifier has already proven to be a robust classifier and successfully used in many applications, it is seldom used jointly with active learning. In this letter, we propose a novel active learning technique for sparse representation classifiers. In the proposed model, the query function is designed by combining uncertainty and diversity criteria, both of which are defined by using the sparse representation classifier in kernel space. The proposed technique outperforms other state-of-the-art methods in terms of classification performance.

General convolutional neural networks (CNNs) process the data in the spatial domain and lack efficacy in modeling the spectral context in multispectral image processing task. To address this limitation, in this letter, we propose a wavelet-inspired attention-based convolution neural network (WIANet) architecture that combines the modeling of the spatio-spectral resolution of multispectral remote sensing (RS) images with a wavelet convolution and attention unit into a single deep learning architecture for land cover classification. Our aim is to add the characteristics of both wavelet transform (WT) and attention mechanism into an UNet-based architecture to better exploit the spectral and texture information for distinguishing classes with high similarity in the spectral signatures. We evaluate the performance of the proposed approach on a multilabel multispectral Sentinel-2 dataset. The experiment shows that the proposed approach performs better than the reference methods in different conditions on the number of training samples.

Deep learning techniques have shown increasing potential in change detection (CD) in hyperspectral images (HSIs). However, most deep learning-based existing methods for HSI CD follow a patch-based local learning framework and concentrate on binary CD. In this letter, we propose an end-to-end semantic feature fused global learning (SFFGL) framework for HSI multiclass change detection (MCD). In SFFGL, a global spatial-wise fully convolutional network (FCN), which introduces a spatial attention mechanism (PAM) between encoder and decoder, is designed to effectively exploit the global spatial information from the whole HSIs and achieve patch-free inference. PAM can adaptively extract global spatial-wise feature representation. In the model training stage, a global hierarchical (GH) sampling strategy is introduced to obtain diverse gradients during backpropagation for more robust performance. The semantic-spatial feature fusion (S2F2) unit is designed to effectively fuse the enhanced spatial context information in the encoder and the semantic information in the decoder. More importantly, a semantic feature enhancement module (SEM) is proposed to weaken the influence of the unchanged regional background on the change regional foreground, thus further improving the accuracy. Experimental results on two benchmark HSI datasets demonstrate the effectiveness of the proposed SFFGL.

In recent years, numerous deep learning (DL)-based frameworks have been proposed for hyperspectral image classification (HSIC). Considering a large number of spectral bands of hyperspectral images (HSIs), it is still challenging to effectively utilize the spectral information and achieve accurate classification when few training samples are available. To make full use of the spectral-spatial information in HSIs with few training samples, in this letter we propose a lightweight end-to-end attention-enhanced feature fusion network (AeF2N). The proposed AeF2N consists of four sequential stages, i.e., spectral feature augmentation, spatial contextual feature interaction, spectral feature augmentation, and classification. The first and third stages are used to capture and augment the discriminative spectral features, while the second stage is used to capture spatial information. Notably, two novel attention blocks, spectral augmentation attention (SAA) and spatial integration attention (SIA) are interactively introduced to capture significant spectral and spatial information, respectively. Based on the proposed spectral and spatial feature discrimination stages, the AeF2N effectively identifies both spectrally significant (e.g., irregular small objects) and spatially significant (e.g., specific-shaped objects) land objects with high accuracy. Experimental results obtained on three benchmark hyperspectral datasets demonstrate the superiority of the proposed approach compared with six state-of-the-art DL-based methods in terms of higher classification accuracy and efficiency.

Lava tubes are terrestrial tunnel-like natural subsurface caves. Mounting evidence suggests their presence on the Moon and Mars. Planetary radar sounders are nadir-looking instruments operating in the high-frequency (HF)/very-HF (VHF) part of the spectrum with subsurface penetration capabilities. Recently, several studies either proposed future mission concepts for lava tubes’ detection or attempted to locate them on the Moon and Mars with the available radar-sounding data. Lava tubes are typically modeled as quasi-cylindrical structures but their actual geometry and their influence on the radar backscattering in near-nadir regime have never been investigated in the literature. These are crucial information for understanding the feasibility of detecting lava tubes by current and future planetary radar sounding systems. Accordingly, in this article: 1) we assess whether lava tubes are self-affine fractal surfaces at horizontal scales relevant to radio and microwave scattering and 2) we evaluate the effect of lava tube topography on the radar backscattering response in the near-nadir regime. Our experimental results, which are inferred from 3-D terrestrial laser scanning (TLS) data of planetary lava tube analogs, show that lava tubes: 1) are self-affine fractals at horizontal scales relevant to radar sounding and 2) they are electromagnetically rough surfaces, especially in the VHF band. We provide quantitative values on the lava tube fractal parameters and radar roughness losses along with a discussion on both: 1) the implication of our results on current radar sounding systems’ ability to detect lava tubes and 2) the planning of future missions devoted to lava tube detection and characterization.

Caves are one of the last frontiers of human exploration on Earth. They are very relevant scientific targets as they host significant biodiversity and unique geologic formations. The presence of underground passages accessible for human or robotic exploration are revealed by localized collapse of the near-surface ceiling of a cave system (skylight). Remote sensing systems are a valuable tool for skylights detection as these features are often located on very remote and often inaccessible regions of the Earth. However, with the available remote sensing techniques and data analysis methodologies, it is very difficult to determine whether a skylight is providing access to a cave continuation or it represents only a closed depression with no extensions. In this article we propose a methodology, based on very high-resolution (VHR) orbital synthetic aperture radar (SAR) imaging systems, to estimate both caves geometric characteristics and accessibility information in the proximity of a skylight. To test our methodology, we acquired radar data over different Earth’s location by exploiting the Capella Space X-band microsatellite radar constellation. The experimental results show that our methodology effectively determines the caves geometric characteristics and accessibility under a variety of surface conditions. We also detected several unknown and unexplored large cave systems located near Volcan Wolf and Ecuador, Isla Isabela, Galapagos. The presented work has relevant implications for the field of geological studies, ecology, and space exploration research since optical imaging shows the evidence of potential cave systems accessible from skylights on other planetary bodies such as Mars.

Semantic segmentation is one of the most challenging tasks for very high resolution (VHR) remote sensing applications. Deep convolutional neural networks (CNN) based on the attention mechanism have shown outstanding performance in VHR remote sensing images semantic segmentation. However, existing attention-guided methods require the estimation of a large number of parameters that are affected by the limited number of available labeled samples that results in underperforming segmentation results. In this paper, we propose a multi-scale feature fusion lightweight model (MSFFL) to greatly reduce the number of parameters and improve the accuracy of semantic segmentation. In this model, two parallel enhanced attention modules, i.e., the spatial attention module (SAM) and the channel attention module (CAM) are designed by introducing encoding position information. Then a covariance calculation strategy is adopted to recalibrate the generated attention maps. The integration of enhanced attention modules into the proposed lightweight module results in an efficient lightweight attention network (LiANet). The performance of the proposed LiANet is assessed on two benchmark datasets. Experimental results demonstrate that LiANet can achieve promising performance with a small number of parameters.

With the increasing number of radar sounder (RS) instruments being used in planetary exploration, there is an increasing need for advanced and efficient RS data simulators. In this context, it is important to combine the advantages of the different simulators to produce end-to-end simulations at multiple scales in a reasonable time. This paper addresses this problem by presenting a novel dictionary-based integrated simulation approach to model both large and small-scale surface scattering phenomena at relatively low computational costs. The method combines the advantages of a ray-tracing approach for simulating large surface areas at low resolution and a numerical technique for simulating the small-scale wave-target interaction at higher resolution. The proposed approach generates an instrument dictionary, (i.e. a finite set of high resolution rough responses at the small-scale) that can be used for surface target simulations. The method is validated by comparing its response with that obtained by a numerical simulator (known to have high accuracy). Then it is demonstrated for the forward modeling of small-scale roughness on a synthetic target and for the inversion of small-scale roughness from existing RS data. The results demonstrate the capability of the method to achieve high accuracy as well as computational efficiency in addressing the problem of small-scale surface scattering on a large-scale scenario.

Recently, pan-sharpening methods based on deep learning (DL) have achieved state-of-the-art results. However, current existing DL-based pan-sharpening methods need to be trained repetitively for different satellite sensors to obtain satisfactory fusion performance and therefore require a large number of training images for each satellite. To deal with these issues, in this paper we propose a unified two-stage spatial and spectral network (UTSN) for pan-sharpening. A branch of networks is constructed for each different satellite, in which the spatial enhancement network (SEN) is shared to improve the spatial details in the fused images from different satellites. A spectral adjustment network (SAN) is employed to capture the spectral characteristics of the specific satellite. Through SAN, the spectral information in the intermediate image from SEN is refined to produce the final fusion results. Such a framework can integrate the datasets from different satellites together for sufficient training of SEN. The proposed method is able to achieve promising pan-sharpening results also for a new satellite with limited training images by only learning a new SAN on the few-shot datasets due to the simple but efficient structure of SAN. The experimental results show that the proposed method can produce state-of-the-art fusion results in both the standard and few-shot cases. The source code is publicly available at https://github.com/RSMagneto/UTSN.

Transformer is a powerful tool for capturing long-range dependencies and has shown impressive performance in hyperspectral image (HSI) classification. However, such power comes with a heavy memory footprint and huge computation burden. In this paper, we propose two types of lightweight self-attention modules (a channel lightweight multi-head self-attention module and a position lightweight multi-head self-attention module) to reduce both memory and computation while associating each pixel or channel with global information. Moreover, we discover that transformers are ineffective in explicitly extracting local and multi-scale features due to the fixed input size and tend to overfit when dealing with a small number of training samples. Therefore, a lightweight transformer (LiT) network, built with the proposed lightweight self-attention modules, is presented. LiT adopts convolutional blocks to explicitly extract local information in early layers and employs transformers to capture long-range dependencies in deep layers. Furthermore, we design a controlled multi-class stratified sampling strategy to generate appropriately sized input data, ensure balanced sampling, and reduce the overlap of feature extraction regions between training and test samples. With appropriate training data, convolutional tokenization, and lightweight transformers, LiT mitigates overfitting and enjoys both high computational efficiency and good performance. Experimental results on several HSI datasets verify the effectiveness of our design.

Deep unfolding networks have obtained satisfactory performance in the pansharpening task owing to their sufficient interpretability. Inspired by the back-projection (BP) mechanism, we propose a BP-driven model, spatial-spectral dual back-project network (S2DBPN), to fuse the low spatial resolution multispectral (LR MS) and the high spatial resolution panchromatic (PAN) images by exploiting the BP in spatial and spectral domains. Specifically, the proposed S2DBPN is made up of a spatial BP network, a spectral BP network, and a reconstruction network. In the spatial BP network, spatial down- and up-projection modules are derived from BP, which is responsible for the projection of the LR MS image into the spatial domain. By analogy with the spatial BP, we reformulate the degradation between high spatial resolution multispectral (HR MS) and PAN images as spectral down- and up-projections. Then, the spectral BP network is constructed for the projection of the PAN image along the channel dimension. Finally, the features from spatial and spectral BP networks are integrated to produce the desired HR MS image through the reconstruction network. Compared to the state-of-the-art methods, extensive experiments on QuickBird, GeoEye-1, and WorldView-2 datasets demonstrate that our S2DBPN produces better HR MS images in terms of qualitative and quantitative evaluation metrics. The code of S2DBPN is released at: https://github.com/RSMagneto/S2DBPN.

Thanks to their capability of modeling global information, transformers have been recently applied to change detection in remote sensing images. Generally, the changes in terms of shape and appearance of objects lead to relation changes among these objects in multi-temporal images. However, in this context, the attention mechanism in transformers has not been fully explored yet to learn relation changes in the observed scenes. In this paper, we analyze the relation changes in multi-temporal images and propose a cross-temporal difference (CTD) attention to capture these changes efficiently. Through the CTD attention, the changed areas are distinguished better from the unchanged areas. Based on the CTD attention, two CTD-transformer encoders are constructed to extract the features of changed areas from the embedded tokens of multi-temporal images in a cross manner. Then, the extracted features at the coarse scale are further improved to the fine-scale by the corresponding CTD-transformer decoders. In addition, consistency-perception blocks (CPBs) are designed to preserve the structures and contours of changed areas. Finally, all extracted features from multi-temporal images are concatenated to produce the desired change map. Compared to state-of-the-art methods, experimental results on LEVIR-CD, WHU-CD, and CLCD datasets demonstrate that the proposed method produces better performance. The source code is available at https://github.com/RSMagneto/CTD-Former.

With the development of high-resolution satellites, more and more attention has been paid to remote sensing (RS) scene classification. Convolutional neural networks (CNNs), which replace the traditional handcrafted features with a learning-based feature extraction mechanism, are widely used in scene classification. But CNNs are less effective in deriving long-range contextual relations, which limits the further improvement. Visual transformer (VT), an emerging image processing method, provides a new perspective for RS scene classification by directly acquiring long-range features. Although there have been limited works combining CNN and VT through simple concatenation, the collaborations between them are insufficient. To address these issues, we propose a local and long-range collaborative framework (L2RCF). First, we design a dual-stream structure to extract the local and long-range features. Second, a cross-feature calibration (CFC) module is designed for them to improve representation of the fusion features. Then, combining deep supervision (DS) and deep mutual learning (DML), a novel joint loss is proposed to enhance the dual-stream feature extractor and further improve the fused features. Finally, a two-stage semi-supervised training strategy is designed to improve performance with unlabeled samples. To demonstrate the effectiveness of L2RCF, we conducted experiments on three widely used RS scene classification data sets: RSSCN7, AID, and NWPU. The results show that L2RCF performs significantly better compared with some state-of-the-art scene classification methods.

In this article, we present a new pansharpening method, a zero-reference generative adversarial network (ZeRGAN), which fuses low spatial resolution multispectral (LR MS) and high spatial resolution panchromatic (PAN) images. In the proposed method, zero-reference indicates that it does not require paired reduced-scale images or unpaired full-scale images for training. To obtain accurate fusion results, we establish an adversarial game between a set of multiscale generators and their corresponding discriminators. Through multiscale generators, the fused high spatial resolution MS (HR MS) images are progressively produced from LR MS and PAN images, while the discriminators aim to distinguish the differences of spatial information between the HR MS images and the PAN images. In other words, the HR MS images are generated from LR MS and PAN images after the optimization of ZeRGAN. Furthermore, we construct a nonreference loss function, including an adversarial loss, spatial and spectral reconstruction losses, a spatial enhancement loss, and an average constancy loss. Through the minimization of the total loss, the spatial details in the HR MS images can be enhanced efficiently. Extensive experiments are implemented on datasets acquired by different satellites. The results demonstrate that the effectiveness of the proposed method compared with the state-of-the-art methods. The source code is publicly available at https://github.com/RSMagneto/ZeRGAN.

The estimation of forest parameters, such as canopy height model (CHM) and above ground biomass (AGB), is of ut-most importance for forest monitoring, carbon-cycle modelling, disturbance analysis, resource inventorying and natural disaster prevention. In this work, we profit from the most recent advancements in deep learning research to propose a convolutional neural network (CNN) architecture for frequent forest parameter estimation at large scale. Our technique consists of a fully convolutional, multi-modal framework, which works on a single set of complementary multi-spectral and interferometric SAR data, acquired by ESA's Sentinel-2 and DLR's TanDEM-X missions, respectively. The regression performance of our framework has been tested over four tropical forest test sites in Gabon, Africa. The estimation of CHM shows promising early results when compared to state-of-the-art methods and has the advantage of requiring only a single input image pair instead of a longer time-series, as commonly done for state-of-the-art model-based techniques.

Cave systems are one of the last frontiers of human exploration on both Earth and other celestial bodies. In this context, lava tubes are natural subsurface tunnels, which are not visible from the surface, that are ubiquitous on Earth as well as on the Moon and Mars. Skylights are one of the surface evidences of the presence of such conduits in the form of overhanging collapses of the cave ceiling, making these cavities partially observable and potentially accessible. In this paper, we propose a method for imaging and characterizing subsurface structures by spaceborne VHR SAR imaging. We performed several acquisitions over different Earth locations by exploiting Capella's X-band VHR SAR imaging radars in spotlight mode. The obtained results show that the proposed methodology is able to characterize the main geometric parameters of the first section of a lava tube (e.g. width, height) in the surroundings of a skylight and it provides an indication of the actual subsurface accessibility. The proposed methodology has several important implications for exploration and could be applied to different type of cavities other than lava tubes.

Spaceborne radar sounders are nadir-looking sensors devoted to subsurface investigations. The data interpretation of these sensors can be severely hindered by clutter originating from surface off-nadir reflections. Recently, the concept of distributed radar sounding has been proposed for synthesizing a narrow radar antenna beam with clutter suppression capability. The antenna beam is effectively synthesized by deploying an array of orbiting sensors in formation flight. In this paper, we assess the capability of distributed radar sounding to further discriminate clutter from subsurface returns by exploiting Direction of Arrival (DOA) estimation techniques. Accordingly, we propose an approach to design and evaluate the distributed radar sounder DOA estimation performance. The theory is complemented by radar simulations of several acquisitions over Greenland. The simulations confirm that clutter discrimination through DOA estimation is an effective approach for further improving the array capability in disam-biguating subsurface echoes from surface ones.

SAR and optical images provide complementary information on land-cover categories in terms of both spectral signatures and dielectric properties. This paper proposes a new unsupervised land-cover segmentation approach based on contrastive learning and vector quantization that jointly uses SAR and optical images. This approach exploits a pseudo-Siamese network to extract and discriminate features of different categories, where one branch is a ResUnet and the other branch is a gumble-softmax vector quantizer. The core idea is to minimize the contrastive loss between the learned features of the two branches. To segment images, for each pixel the output of gumble-softmax is discretized as a one-hot vector and its proxy label is chosen as the corresponding class. The proposed approach is validated on a subset of DFC2020 dataset including six different land-cover categories. Experimental results demonstrate improvements over the current state-of-the-art techniques and the effectiveness of unsupervised land-cover segmentation on SAR-optical image pairs.

We propose an unsupervised methodology for multi-class change detection (CD) in multimodal remote sensing data fused using the Kronecker product formalism. The method utilizes the compressed change vector analysis (C2VA) on the fully vectorized change matrices. The multimodal case is demonstrated using dual-frequency full-polarimetric Syn-thetic Aperture Radar (SAR) data obtained by EMISAR over the Foulum agricultural area. The change types are inves-tigated using ground truth data for the growth of various crops. The work showcases the capability of the Kronecker product-based CD formalism beyond conventional scalar change indices.

Radar sounders (RSs) are widely used to image profiles (radargrams) of the subsurface of planetary bodies and the Earth. However, despite the huge scientific return from radargram analyses, their horizontal and vertical resolutions are limited by technical factors. Even if methods exist for improving the resolution, these are still limited by technical factors and introduce artifacts. This paper proposes an unsupervised deep-learning method that synthesizes accurate super-resolved radargrams overcoming these limitations. The method adopts the Cycle-Consistent Adversarial Network (CyleGAN) that learns the mapping function between the low- and high-resolution data distributions. The network is adapted to match the low- and high-resolution radargram characteristics, including the differences in dimensions and radiometric properties. The proposed method was successfully validated on airborne data at higher resolution and simulated data with lower resolution.

Lifelong learning is important in remote sensing image understanding, especially in an open world where streaming of data are incrementally available. Current related research mainly focuses on preserving learned knowledge (i.e., avoiding catastrophic forgetting) while devoting less attention to the exploitation of historical knowledge to facilitate the learning of new knowledge. Here, we propose a new framework to bridge this gap. It consists of two sub-networks that memorize old and learn new knowledge separately, and exploit the synergy of transfer cells and triple distillation to take advantage of the valuable knowledge from previous learned tasks to facilitate learning new tasks while avoiding forgetting old tasks. Furthermore, it uses an asymmetric structure considering feature generality on historical tasks and scale- and channel-feature dependence on remote sensing images for specific tasks. Experimental results obtained in scene classification on several open benchmarks demonstrate the effectiveness of the framework.

Ship detection using synthetic aperture radar images is a key technology in maritime surveillance applications. In addition to the position of the vessel, the characterization of the target (length, width and orientation) is often a requirement. In this paper, we present a deep learning architecture for object detection we developed by modifying the popular YOLOv3 architecture to apply to vessel detection and parameter estimation from SAR images. The proposed architecture was trained and tested on a large dataset of SAR images defined in this work. It contains images covering a wide range of spatial resolutions (pixel spacing ranging from 1.5m to 50m) and labelled with oriented bounding boxes to associate to each vessel not only its position but also size and orientation. The obtained results are very promising and confirm the validity of the approach.

Radar sounders (RS) have the unique capability of providing direct measurements of subsurface structures. Due to the complex nature of subsurface acquisition scenarios, the RS design and data interpretation require the support of SAR focused electromagnetic simulations of the subsurface targets. While SAR focusing techniques applied to real RS data are well-established, the same techniques cannot be directly applied to the simulated radargrams, due to the inherent differences between the real data and the data simulation processing chains. In this paper, we approach this challenge by adopting a novel approach to the focusing of simulated raw RS data. In the proposed approach the main focusing parameters are extracted from the simulator impulse response to properly consider any artefact and the inherent assumptions exploited in the simulations. The method has been applied to previously published raw simulated radargrams of two RS instruments. The results show that the method provides good results for different target scenarios and RS instruments.

Availability of limited training remote sensing datasets is one of the problems in deep learning, as deep architectures require a large number of training samples for proper training. In this paper, we present a technique for data augmentation based on a spectral indexed generative adversarial network to train deep convolutional neural networks. This technique uses the spectral characteristic of multispectral (MS) images to support data augmentation in order to generate realistic training samples with respect to each land-use and land-cover class. The impact of multispectral remote sensing data generated through the spectral indexed GAN are evaluated through classification experiments. Experimental results obtained on the classification of the Sentinel-2 Eurosatallband datasets show that data augmentation through spectral indexed GAN enhances the main accuracy metrics.

ESA's EnVision mission has been selected for Venus exploration, with launch scheduled in 2031. It will carryon-board the Subsurface Radar Sounder (SRS) to profile the shallow crust at low frequency and support the understanding of Venus' geological history. In the design and performance assessment of SRS, an important step is the analysis of off-nadir clutter due to rough surface scattering and its potential to mask the scientifically relevant subsurface echoes. While there have been studies to understand the roughness of Venusian terrains from data available from the previous missions, there are few studies that analyze the roughness and clutter at the wavelength-scale of a low-frequency radar sounder instrument. In this paper, we present a first step towards filling this gap by analyzing the clutter performance of selected sites on Venus where subsurface interfaces are expected, namely the plains, the impact craters, and the lava flows. Using surface roughness parameters and fractal modelling, we generate multiple surface realizations, simulate the SRS clutter response and analyze the probability distribution of clutter depth and power. The results show that for most of the sites, clutter is concentrated very close to the surface, and therefore does not significantly affect the subsurface detection.

The very high spectral resolution in hyperspectral images (HSIs) presents an opportunity to detect subtle land-cover changes. However, availability of HSIs acquired from different platforms requires the development of change detection (CD) methods for HSIs capable to process images with different spatial resolutions. In this paper, we propose an end-to-end sub-pixel convolution-based residual network (SPCNet) to detect changes between high resolution (HR) and low resolution (LR) HSIs. First, an efficient sub-pixel convolution layer is introduced to upscale the LR feature maps into the HR one. Then, the super resolution (SR) block is designed to generate more discriminative representations in sub-pixel-based LR images. Moreover, the sub-pixel-based feature of LR image and pixel-based feature of HR image are concatenated as an input to the designed ResNet for HSI CD. Experimental results on two HSI datasets demonstrate the effectiveness of the proposed SPCNet.

There are limited studies on the semantic segmentation of high-resolution polarimetric synthetic aperture radar (PolSAR) images due to the scarcity of training data and the complexity of managing speckle noise. The Gaofen contest has provided open access a high-quality PolSAR semantic segmentation dataset. Taking this opportunity, we propose a multipath residual network (MP-ResNet) architecture for the semantic segmentation of high-resolution PolSAR images. Compared to conventional U-shape encoder–decoder convolutional neural network (CNN) architectures, the MP-ResNet learns semantic context with its parallel multiscale branches, which greatly enlarges its valid receptive fields and improves the embedding of local discriminative features. In addition, MP-ResNet adopts a multilevel feature fusion design in its decoder to effectively exploit the features learned from its different branches. Comparisons with the baseline method of fully connected network (FCN with ResNet34) show that the MP-ResNet has achieved significant accuracy improvements. It also surpasses several state-of-the-art methods in terms of overall accuracy (OA),$\text{m}F_{1}$and frequency weighted intersection over union (fwIoU), with only a limited increase of computational costs. This CNN architecture can be used as a baseline method for future studies on the semantic segmentation of PolSAR images. The code is available at:https://github.com/ggsDing/SARSeg.

Monitoring forest dynamics is of critical importance for both sustainable forest management and conservation purposes. Light detection and ranging (lidar) data provide a detailed representation of the 3-D structure of forest stands that can be used to analyze a number of trees and stand characteristics. Recently, multiple lidar acquisitions over the same area are becoming more common allowing changes in stand attributes to be assessed over time. In order to effectively utilize such multitemporal data sets for forest dynamics monitoring, we propose a method for unsupervised change detection (CD) of lidar data based on polar change vector analysis (CVA). The proposed method involves extracting relevant lidar point cloud metrics for a given area over time. Pixel-wise difference vectors of the metrics are then converted from Cartesian to polar coordinates to represent the magnitude and direction of change. Finally, the change vectors are analyzed in the polar domain to automatically discriminate between the different classes of change. The method is applied to a multitemporal lidar data set of coniferous forest on Vancouver Island, British Columbia, Canada, impacted by various types of land cover change. The experimental results demonstrate that the proposed method is capable of automatically discriminating between different classes of lidar change.

This letter presents a novel approach to tree-top detection in heterogeneous forest structures characterized by mixed species using high-density light detection and ranging (LiDAR) data. Although literature techniques can achieve accurate results in even-size and even-age homogeneous forests, they detect several false tree tops in forests characterized by variable crown dimensions. To solve this problem, the proposed method 1) identifies a preliminary set of candidate tree tops (CTPs) used to build a triangulated network; 2) performs an edge-based local forest analysis to identify groups of CTPs having the highest probability of belonging to the same crown; and 3) removes false tree tops according to a local directed graph analysis. To address large-scale forest analysis, the method exploits the Delaunay triangulation that efficiently defines a network topology made up only by relevant edges, thus sharply reducing the edge-based analyses to be performed. Given the triangulated network properties, the computational effort of the local analysis is not affected by the network size. The method has been tested in a mixed multi-layer multi-age forest located in the southern Italian Alps. The results obtained demonstrate that this computationally scalable algorithm outperforms standard tree-top detection methods increasing the overall detection accuracy up to 15.3%.

This letter proposes a novel method based on Deep Learning (DL) to forest species classification in airborne Light Detection and Ranging (LiDAR) data. Differently from the state-of- the-art approaches, the proposed method: (1) does not assume any prior knowledge either on the forest to be classified or on the sensor used to acquire the LiDAR data, and (2) can be applied to heterogeneous forest characterized by mixed species. First, the 3D point cloud of each individual tree is decomposed into 8 angular sectors to generate a multi-slices representation of the vertical structure of the tree. This representation models the foliage, the stem and the branches of the tree crown as well as depicts the internal and external crown properties. Then, a Multi-View CNN (MVCNN) DL automatically extracts features used to discriminate the different tree species. This network is pre-trained on the massive ImageNet database, thus guaranteeing fast convergence with a relatively small number of ground reference data. Experiments were carried out on high density airborne LiDAR data collected over a multi-layer multi-age forest characterized by four conifers and three broadleaf species. The proposed method outperformed the state-of-the-art approaches increasing the Overall Accuracy (OA) up to 16% and 18.9% compared to a DL and a shallow tree species classification methods, respectively. When applied to coniferous or broadlaef forests, the proposed method showed an increase of OA 10.1% and 15.9% (for conifers), and 9.5% and 21.6% (for broadleafs) compared to the DL and shallow methods, respectively.

Remote sensing (RS) scene classification has attracted extensive attention due to its large number of applications. Recently, convolutional neural networks (CNNs) methods have shown impressive ability of feature learning in RS scene classification. However, the performance is still limited by large-scale variance and complex background. To address these problems, we present a multilayer feature fusion network with spatial attention and gated mechanism (MLF2Net_SAGM) for RS scene classification. At first, the backbone is employed to extract multilayer convolutional features. Then, a residual spatial attention module (RSAM) is proposed to enhance discriminative regions of the multilayer feature maps, and key areas can be harvested. Finally, the multilayer spatial calibration features are fused to form the final feature map, and a gated fusion module (GFM) is designed to eliminate feature redundancy and mutual exclusion (FRME). To verify the effectiveness of the proposed method, we conduct comparative experiments based on three widely used RS image scene classification benchmarks. The results show that the direct fusion of multilayer features via element-wise addition leads to FRME, whereas our method fuses multilayer features more effectively and improves the performance of scene classification.

Supervised classification of remote sensing data requires a large number of high-quality annotated samples. At the operational level, the definition of a large training set by photograph interpretation is costly and time-consuming. The manual annotation activity is typically supported by high-resolution satellite data. Therefore, when working at country or continental scale, it is necessary to efficiently access large archives of remotely sensed data. To address these issues, this letter presents an interactive strategy implemented in a cloud-computing platform for defining effective training sets with significantly reduced human effort. This is achieved by combining active learning (AL) and self-paced learning (SPL) techniques. First, an initial training set is used to classify the pool of unlabeled samples. Then, the method progressively adds high-confidence samples, selected through an SPL strategy, and low-confidence samples selected considering an AL strategy. While the high-confidence sample labels are self-paced, the low-confidence ones are manually assigned. The cloud-computing platform allows the: 1) definition of a complete training set in a fast and efficient way and 2) access to a multipetabyte catalog of satellite imagery. Experiments carried out on the Google Earth Engine (GEE) Platform demonstrate the effectiveness of the proposed strategy compared to the standard manual annotation.

Deep learning has attracted intensive attention in synthetic aperture radar (SAR) automatic target recognition (ATR). Usually, a considerable number of labeled samples are necessary to learn a deep model for obtaining good generalization capability. However, the process of sample labeling is time-consuming and costly. This letter proposes an active self-paced deep learning (ASPDL) approach to SAR ATR. In a nutshell, we first introduce the Bayesian inference into the process of deep model parameter optimization, aiming at learning a robust classification model in the case of a limited number of labeled samples. Next, a cost-effective sample selection strategy is presented to iteratively and actively select the informative samples from a pool of unlabeled samples for labeling. Concretely, high-confidence samples are actively selected through self-paced learning (SPL) way and automatically pseudo-labeled with the current classification model, whereas low-confidence samples are chosen through an active learning strategy and manually labeled. Finally, we update the parameters of the model by minimizing a dual-loss function using a new training set that is constructed by incorporating new labeled samples with original ones. Experiments on the moving and stationary target acquisition and recognition (MSTAR) benchmark data demonstrate that the proposed method can achieve better classification accuracy with relatively few labeled samples compared with some state-of-the-art methods.

Deep learning-based unsupervised change detection (CD) methods compare a prechange and a postchange image in deep feature space and require precise knowledge of the event date for selecting proper pre-/post-change images. However, in many applications changes may occur gradually over a span of time making pre-/post-dates difficult to establish or prior knowledge of event date is unknown. On the other hand, deep learning-based time-series analysis methods are generally supervised. Considering such scenarios, we propose a novel unsupervised deep learning-based method to detect changes in an image time-series. The method does not make any assumption on the date of the occurrence of the change event. It treats CD as an anomaly detection problem by exploiting multilayer long short term memory (LSTM) network to learn a representation of the time series. The proposed method ingests a shuffled time series and uses an encoder–decoder LSTM model to rearrange the input sequence in correct order. While the model fails to rearrange the changed pixels, unchanged data can be rearranged in the correct order. This enables the identification of the changed pixels. To show the effectiveness of the proposed method, we tested it on two multitemporal Sentinel-1 data sets over Brumadinho, Brazil, and Bhavanisagar, India.

Generative models are typically employed to approximate the distribution of deep features. Recently, these state-of-the-art methods have been applied to estimate image transformations by an unsupervised learning approach. In this letter, a novel spectral index generative adversarial network (SIGAN) is proposed for the generation of multispectral (MS) remote sensing images. This network is defined to effectively perform data augmentation starting from a limited number of training samples in the MS remote sensing domain for training deep learning models. The SIGAN model is able to capture class-specific properties in data augmentation, by incorporating the task-specific normalized spectral indices to model class-by-class properties of MS images. Experimental results obtained on a Sentinel 2 dataset show that the proposed model provides better performance than other generative adversarial networks (GANs) in MS data generation.

Deep Convolutional Neural Networks are state-of-the-art methods in the domain of classification of remote sensing (RS) data. However, traditional CNN models suffer from huge computational costs in learning land-use and land-cover features, particularly in large scale RS problems. To address this issue, we propose a reliable mono and dual Regulated contractive-expansive-contractive (CEC) convolutional neural network for scene based multispectral (MS) image classification. The proposed technique increases the accuracy of learning and minimizes the loss in the feature maps by incorporating the CEC approach in the classification. Extensive experiments conducted on the Sentinel-2 EuroSATallbands dataset pointed out that the proposed model outperforms state-of-the-art models such as EfficientNet-B0, RESNet-50 and EfficientNet-B7.

Several studies are in progress for proposing an Earth orbiting radar sounder (EORS) mission. Some of them consider as baseline system architecture a recently proposed distributed radar sounding array in formation flight with enhanced capabilities of clutter suppression. Besides clutter, the detectability of subsurface targets may also be affected by the propagation of the radar signal through Earth’s ionosphere. These effects include frequency-dependent phase dispersion and scintillations. In this letter, we present a subsurface detection performance assessment of an EORS with distributed architecture focusing on the ionospheric effects. The novel contributions of this work are: (i) simulation of the coherent radar response of a representative polar ice target (englacial layering) in the distributed radar sounding case; (ii) inclusion of spatially-dependent ionospheric scintillation effects on the distributed beam pattern; (iii) inclusion of phase dispersion effects for different values of total electron content (TEC), and (iv) analysis of the subsurface detection performance. Detectability analysis is performed after applying a state-of-the-art technique for compensating ionospheric phase-dispersion effects. The results show that the englacial layering is detectable by compensating the dispersion effects in the range between 1 and 21 TECU in the ionosphere. The layering is also detectable at higher values of TEC by improving the accuracy of TEC estimation. Moreover, even without compensation, the worst-case ionospheric phase scintillations of 25° produces a negligible effect on the detectability.

Pansharpening methods based on deep neural networks (DNNs) have been attracting great attention due to their powerful representation capabilities. In this article, to combine the feature maps from different subnetworks efficiently, we propose a novel pansharpening method based on a spatial and spectral extraction network (SSE-Net). Different from the other methods based on DNNs that directly concatenate the features from different subnetworks, we design adaptive feature fusion modules (AFFMs) to merge these features according to their information content. First, the spatial and spectral features are extracted by the subnetworks from low spatial resolution multispectral (LR MS) and panchromatic (PAN) images. Then, by fusing the features at different levels, the desired high spatial resolution MS (HR MS) images are generated by the fusion network consisting of AFFMs. In the fusion network, the features from different subnetworks are integrated adaptively, and the redundancy among them is reduced. Moreover, the spectral ratio loss and the gradient loss are defined to ensure the effective learning of spatial and spectral features. The spectral ratio loss captures the nonlinear relationships among the bands in the MS image to reduce the spectral distortions in the fusion result. Extensive experiments were conducted on QuickBird and GeoEye-1 satellite datasets. Visual and numerical results demonstrate that the proposed method produces better fusion results compared with literature techniques. The source code is available athttps://github.com/RSMagneto/SSE-Net.

The use of deep learning (DL) methods for change detection (CD) is currently dominated by supervised models that require a large number of labeled samples. However, these samples are difficult to acquire in the multitemporal case. A possible alternative is leveraging methods that exploit transfer learning for CD by reusing DL models pretrained for other tasks. However, the performance of the transfer-learning-based models decreases as much as the target images differ from the ones used for training the model. To overcome this limit, we propose an unsupervised CD method that exploits multiresolution deep feature maps derived by a convolutional autoencoder (CAE). It automatically learns spatial features from the input during the training phase without requiring any labeled data. The proposed method processes the bitemporal images to obtain and compare multiresolution bitemporal feature maps. These feature maps are then analyzed by a feature-selection technique to select the most discriminant ones. Furthermore, an aggregated multiresolution difference image is computed and used for a detail-preserving multiscale CD. In the context of this CD approach, we propose two alternative strategies to retrieve multiscale reliability maps. We tested the proposed method on bitemporal multispectral images acquired by Landsat-5 and Landsat-8 representing burned areas and Sentinel-2 images representing deforested areas. Results confirm the effectiveness of the proposed CD technique.

Research based on Mars Advanced Radar for Subsurface and Ionosphere Sounding (MARSIS) data detected unusual radar bright basal reflections located at about 1.5 km depth in a Mars region denoted as Ultimi Scopuli. These reflections were interpreted as a signature of subglacial liquid water even though this interpretation is still being debated in the literature. In this article, we propose a novel approach to the detection and imaging of candidate subglacial liquid water from radar sounding data. The approach combines the radar echo power traces with a suitable digital elevation model to provide a bidimensional representation of the surface. Even if the imaging method reconstructs a representation of the surface, we prove that it can be used to identify subsurface bright reflections in icy regions. Imaging is feasible even if the basal interface is not directly included in the processed data for image generation. To support this experimental evidence, we show that a relationship exists between the value of the reflected echo power originating from the englacial layers and the basal-to-surface-echo-power ratio. The observed relationship holds on both Ultimi Scopuli radar sounding data acquired on Mars and Lake Vostok data acquired on Earth. Our results show that the 2-D imaging provides an alternative way for locating candidate subglacial liquid water bodies on Mars over large areas also where the basal interface is not directly measured. The proposed approach complements previous research for further evaluation of the actual presence of liquid water on Mars.

Spaceborne radar sounders are High Frequency (HF) / Very High Frequency (VHF) nadir-looking sensors devoted to subsurface investigations. Their data interpretation can be severely hindered by off-nadir surface clutter. Recent literature showed that the clutter suppression capabilities of this class of systems can be greatly enhanced by deploying an array of orbiting sensors in formation flight synthesizing a narrow radar antenna beam. In this paper, we assess the capability of distributed radar sounding to discriminate clutter from subsurface returns by exploiting Direction of Arrival (DOA) estimation techniques. This is achieved by first outlining an approach for designing and evaluating the distributed radar sounder DOA estimation performance as function of the radar system parameters (e.g, inter-sensor distance) and external noise factors such as ionospheric scintillations. Then, the theory is complemented by radar simulations of several acquisitions over Greenland assuming a variety of subsurface geometries. The simulations confirm that clutter discrimination through DOA estimation is a viable approach to further improve the array capability in disambiguation of subsurface echoes from surface ones.

The large amount of unlabeled remote sensing acquired from different sources and at different times (defined as multiple views in this article) presents both an opportunity and a challenge for change detection. Recently, many generative model-based methods have been proposed for remote sensing image change detection on such unlabeled data. However, the high diversities in the learned features weaken the discrimination of the relevant change indicators in unsupervised change detection tasks. Moreover, these methods lack research on massive archived images. In this work, a self-supervised change detection approach based on an unlabeled multiview setting is proposed to overcome this limitation. This is achieved by the use of a multiview contrastive loss in the feature alignment between multiview images. In this approach, a pseudo-Siamese network is trained to regress the output between its two branches pretrained in a contrastive way on a large dataset of single-sensor or cross-sensor image pairs. Finally, the feature distance between the outputs of the two branches is used to define a change measure, which can be analyzed by thresholding to get the final binary change map. Experiments are carried out on two single-sensor and three cross-sensor datasets. The proposed approach is compared with other supervised and unsupervised state-of-the-art change detection methods. Results demonstrate both improvements over state-of-the-art unsupervised methods and the proposed approach narrows the gap between unsupervised and supervised change detection.

The effective combination of the complementary information provided by huge amount of unlabeled multisensor data (e.g., synthetic aperture radar (SAR) and optical images) is a critical issue in remote sensing. Recently, contrastive learning methods have reached remarkable success in obtaining meaningful feature representations from multiview data. However, these methods only focus on image-level features, which may not satisfy the requirement for dense prediction tasks such as land-cover mapping. In this work, we propose a self-supervised framework for SAR-optical data fusion and land-cover mapping tasks. SAR and optical images are fused by using a multiview contrastive loss at image level and super-pixel level according to one of those possible strategies: in the early, intermediate, and late strategies. For the land-cover mapping task, we assign each pixel a land-cover class by the joint use of pretrained features and spectral information of the image itself. Experimental results show that the proposed approach not only achieves a comparable accuracy but also reduces the dimension of features with respect to the image-level contrastive learning method. Among three fusion strategies, the intermediate fusion strategy achieves the best performance. The combination of the pixel-level fusion approach and the self-training on spectral indices leads to further improvements in the land-cover mapping task with respect to the image-level fusion approach, especially with sparse pseudo labels. The code to reproduce our results will be found athttps://github.com/yusin2it/SARoptical_fusion.

Deep-learning techniques have achieved great success in remote-sensing image change detection. Most of them are supervised techniques, which usually require large amounts of training data and are limited to a particular application. Self-supervised methods solve these problems and are widely used in unsupervised binary change detection tasks. However, the existing self-supervised methods in change detection are suboptimal for pixel-wise change detection tasks. In this work, a pixel-wise contrastive approach is proposed to overcome this limitation. This is achieved by using contrastive loss in superpixel-level features on an unlabeled multiview setting. In this approach, a pseudo-Siamese network is trained to obtain pixel-wise representations and to align features from shifted image pairs. The final binary change map is obtained by using thresholding methods on learned temporal features. To overcome the season-related noise in binary change maps, we also used an uncertainty method to enhance the temporal robustness of the proposed approach. Two homogeneous (OSCD and MUDS) datasets and one heterogeneous (California Flood) dataset are used to evaluate the performance of the proposed approach. Results demonstrate improvements in both efficiency and accuracy over the patch-wise multiview contrastive method.

Semantic change detection (SCD) extends the multiclass change detection (MCD) task to provide not only the change locations but also the detailed land-cover/land-use (LCLU) categories before and after the observation intervals. This fine-grained semantic change information is very useful in many applications. Recent studies indicate that the SCD can be modeled through a triple-branch convolutional neural network (CNN), which contains two temporal branches and a change branch. However, in this architecture, the communications between the temporal branches and the change branch are insufficient. To overcome the limitations in existing methods, we propose a novel CNN architecture for the SCD, where the semantic temporal features are merged in a deep CD unit. Furthermore, we elaborate on this architecture to reason the bi-temporal semantic correlations. The resulting bi-temporal semantic reasoning network (Bi-SRNet) contains two types of semantic reasoning blocks to reason both single-temporal and cross-temporal semantic correlations, as well as a novel loss function to improve the semantic consistency of change detection results. Experimental results on a benchmark dataset show that the proposed architecture obtains significant accuracy improvements over the existing approaches, while the added designs in the Bi-SRNet further improve the segmentation of both semantic categories and the changed areas. The codes in this article are accessible athttps://github.com/ggsDing/Bi-SRNet.

Long-range contextual information is crucial for the semantic segmentation of high-resolution (HR) remote sensing images (RSIs). However, image cropping operations, commonly used for training neural networks, limit the perception of long-range contexts in large RSIs. To overcome this limitation, we propose a wide-context network (WiCoNet) for the semantic segmentation of HR RSIs. Apart from extracting local features with a conventional convolutional neural network (CNN), the WiCoNet has an extra context branch to aggregate information from a larger image area. Moreover, we introduce a context transformer to embed contextual information from the context branch and selectively project it onto the local features. The context transformer extends the vision transformer, an emerging kind of neural networks, to model the dual-branch semantic correlations. It overcomes the locality limitation of CNNs and enables the WiCoNet to see the bigger picture before segmenting the land-cover/land-use (LCLU) classes. Ablation studies and comparative experiments conducted on several benchmark datasets demonstrate the effectiveness of the proposed method. In addition, we present a new Beijing Land-Use (BLU) dataset. This is a large-scale HR satellite dataset with high-quality and fine-grained reference labels, which can facilitate future studies in this field.

During the last decades, radar sounders provided direct measurements (radargrams) of the Earth’s polar caps’ subsurface. Radargrams are of critical importance for a better understanding of glaciologic structures and processes of the ice sheet in the framework of climate change. This article aims to automatically extract information on basal boundary conditions given their substantial relevance for modeling the ice-sheet processes, such as the sliding. We introduce a novel automatic method based on deep learning to detect the basal layer and basal units in radargrams acquired in the inland of icy areas. Radargrams are segmented into englacial layers, bedrock, basal units, and noise-limited regions; the latter includes the echo-free zone (EFZ), thermal noise, and signal perturbation. The network is a U-Net with attention gates and the Atrous Spatial Pyramid Pooling (ASPP) module that automatically extract semantically meaningful features at different scales. Experimental results on two datasets acquired in north Greenland and west Antarctica by the Multichannel Coherent Radar Depth Sounder (MCoRDS3) indicate a high overall segmentation accuracy. The accuracy of basal ice and signal perturbation detection is high, and that of the other classes is comparable with the literature techniques based on handcrafted features. The results show the effectiveness of the proposed method in automatically extracting semantically meaningful features to segment radargrams and map the basal layer and basal units.

Lava tubes are buried channels that transport thermally insulated lava. Nowadays, lava tubes on the Moon are believed to be empty and thus indicated as potential habitats for humankind. In recent years, several studies investigated possible lava tube locations, considering the gravity anomaly distribution and surficial volcanic features. This article proposes a novel and unsupervised method to map candidate buried empty lava tubes in radar sounder data (radargrams) and extract their physical properties. The approach relies on a model that describes the geometrical and electromagnetic (EM) properties of lava tubes in radargrams. According to this model, reflections in radargrams are automatically detected and analyzed with a fuzzy system to identify those associated with lava tube boundaries and reject the others. The fuzzy rules consider the EM and geometrical properties of lava tubes, and thus, their appearance in radargrams. The proposed method can address the complex task of identifying candidate lava tubes on a large number of radargrams in an automatic, fast, and objective way. The final decision on candidate lava tubes should be taken in postprocessing by expert planetologists. The proposed method is tested on both a real and a simulated data set of radargrams acquired on the Moon by the Lunar Radar Sounder (LRS). Identified candidate lava tubes are processed to extract geometrical parameters, such as the depth and the thickness of the crust (roof).

Ship detection in optical remote sensing images plays a significant role in a wide range of civilian and military tasks. However, it is still a challenging issue owing to complex environmental interferences and a large variety of target scales and positions. To overcome these limitations, we propose a supervised multi-scale attention-guided detection framework, which can effectively detect ships of different scales both in complex pure ocean and port scenes. Specifically, a multi-scale supervision module is first proposed to adjust the semantic consistency of different feature levels, obtaining extracted features with small semantic gaps. Next, an attention-guided module is utilized to aggregate context information from both spatial and channel dimensions by calculating map correlations, adaptively enhancing the feature representation. Moreover, to preserve the attribute and spatial relationship of the optimized features, we adopt a capsule-based module as the classifier and obtain satisfactory classification performance. Experimental results conducted on two public high-quality datasets demonstrate that the proposed method obtains state-of-the-art performance in comparison with several advanced methods.

With the increasing availability and resolution of satellite sensor data, multispectral (MS) and panchromatic (PAN) images are the most popular data that are used in remote sensing among applications. This article proposes a novel cross-resolution hidden layer feature fusion (CRHFF) approach for joint classification of multiresolution MS and PAN images. In particular, shallow spectral and spatial features at a global scale are first extracted from an MS image. Then, deep cross-resolution hidden layer features extracted from MS and PAN are fused from patches at a local scale according to an autoencoder (AE)-like deep network. Finally, the selected multiresolution hidden layer features are classified in a supervised manner. By taking advantage of integrated shallow-to-deep and global-to-local features from the high-resolution MS and PAN images, the cross-resolution latent information can be extracted and fused in order to better model imaged objects from the multimodal representation and finally increase the classification accuracy. Experimental results obtained on three real multiresolution datasets covering complex urban scenarios confirm the effectiveness of the proposed approach in terms of higher accuracy and robustness with respect to literature methods.

With more detailed spatial information being represented in very-high-resolution (VHR) remote sensing images, stringent requirements are imposed on accurate image classification. Due to the diverse land-objects with intraclass variation and interclass similarity, efficient and fine classification of VHR images especially in complex scenes is challenging. Even for some popular deep learning (DL) frameworks, geometric details of land-object may be lost in deep feature levels, so it is difficult to maintain the highly-detailed spatial information (e.g., edges, small objects) only relying on the last high-level layer. Moreover, many of the newly developed DL methods require massive well-labeled samples, which inevitably deteriorates the model generalization ability under the few-shot learning. Therefore, in this paper, a lightweight shallow-to-deep feature fusion network (SDF2N) is proposed for VHR image classification, where the traditional machine learning (ML) and DL schemes are integrated to learn rich and representative information to improve the classification accuracy. In particular, the shallow spectral-spatial features are first extracted, and then a novel triple-stage fusion (TSF) module is designed to learn the saliency and discriminative information at different levels for classification. The TSF module includes three feature fusion stages, i.e., low-level spectral-spatial feature fusion, middle-level multi-scale feature fusion, and high-level multi-layer feature fusion. The proposed SDF2N takes advantages of the shallow-to-deep features, which can extract representative and complementary information of crossing layers. It is important to note that even with limited training samples, the SDF2N still can achieve satisfying classification performance. Experimental results obtained on three real VHR remote sensing data sets including two multispectral and one airborne hyperspectral images covering complex urban scenarios confirm the effectiveness of the proposed approach compared with the state-of-the-art methods.

Change detection is a long-standing and challenging problem in remote sensing. Very often, features about changes are difficult to model beforehand, thus making the collection of changed samples a challenging task. In comparison, it is much easier to collect numerous no-change samples. It is possible to define a change detection approach by using only easily available annotated no-change samples, which we henceforth call one-class change detection. Autoencoder networks being trained on no-change data are natural candidates for addressing this task due to their superior performance as compared to other one-class classification models. However, the autoencoders usually suffer from the problem of overgeneralization, i.e., they tend to generalize too well, thus risking properly reconstructing changed samples. In this paper, we propose a novel data-enclosing-ball minimizing autoencoder (DebM-AE) that is trained with dual objectives—a reconstruction error criterion and a minimum volume criterion. The network learns a compact latent space, where encodings of no-change samples have low intra-class variance, which as counter part has the identification of changed instances. We conducted extensive experiments on three real-world data sets. Results demonstrate advantages of the proposed method over other competitors. We make our data and code publicly available1.

Band selection refers to the process of choosing the most relevant bands in a hyperspectral image. By selecting a limited number of optimal bands, we aim at speeding up model training, improving accuracy, or both. It reduces redundancy among spectral bands while trying to preserve the original information of the image. By now, many efforts have been made to develop unsupervised band selection approaches, of which the majorities are heuristic algorithms devised by trial and error. In this article, we are interested in training an intelligent agent that, given a hyperspectral image, is capable of automatically learning policy to select an optimal band subset without any hand-engineered reasoning. To this end, we frame the problem of unsupervised band selection as a Markov decision process, propose an effective method to parameterize it, and finally solve the problem by deep reinforcement learning. Once the agent is trained, it learns a band-selection policy that guides the agent to sequentially select bands by fully exploiting the hyperspectral image and previously picked bands. Furthermore, we propose two different reward schemes for the environment simulation of deep reinforcement learning and compare them in experiments. This, to the best of our knowledge, is the first study that explores a deep reinforcement learning model for hyperspectral image analysis, thus opening a new door for future research and showcasing the great potential of deep reinforcement learning in remote sensing applications. Extensive experiments are carried out on four hyperspectral data sets, and experimental results demonstrate the effectiveness of the proposed method. The code is publicly available.

Convolutional neural networks (CNNs) have achieved tremendous success in computer vision tasks, such as building extraction. However, due to domain shift, the performance of the CNNs drops sharply on unseen data from another domain, leading to poor generalization. As it is costly and time-consuming to acquire dense annotations for remote-sensing (RS) images, developing algorithms that can transfer knowledge from a labeled source domain to an unlabeled target domain is of great significance. To this end, we propose a novel full-level domain adaptation network (FDANet) for building extraction by combining image-, feature-, and output-level information effectively. At the input level, a simple Wallis filter method is employed to transfer source images into target-like ones whereby alleviating radiometric discrepancy and achieving image-level alignment. To further reduce domain shift, adversarial learning is used to enforce feature distribution consistency constraints between the source and target images. In this way, feature-level alignment can be embedded effectively. At the output level, a mean-teacher model is introduced to enforce transformation-consistent constraint for the target output so that the regularization effect is enhanced and the uncertain predictions can be suppressed as much as possible. To further improve the performance, a novel self-training strategy is also employed by using pseudo labels. The effectiveness of the proposed FDANet is verified on three diverse high-resolution aerial datasets with different resolutions and scenarios. Extensive experimental results and ablation studies demonstrated the superiority of the proposed method.

Deep learning has gained broad interest in remote sensing image scene classification thanks to the effectiveness of deep neural networks in extracting the semantics from complex data. However, deep networks require large amounts of training samples to obtain good generalization capabilities and are sensitive to errors in the training labels. This is a problem in remote sensing since highly reliable labels can be obtained at high costs and in limited amount. However, many sources of less reliable labeled data are available, e.g., obsolete digital maps. In order to train deep networks with larger datasets, we propose both the combination of single or multiple weak sources of labeled data with a small but reliable dataset to generate multisource labeled datasets and a novel training strategy where the reliability of each source is taken into consideration. This is done by exploiting the transition matrices describing the statistics of the errors of each source. The transition matrices are embedded into the labels and used during the training process to weigh each label according to the related source. The proposed method acts as a weighting scheme at gradient level, where each instance contributes with different weights to the optimization of different classes. The effectiveness of the proposed method is validated by experiments on different datasets. The results proved the robustness and capability of leveraging on unreliable source of labels of the proposed method.

Sparse unmixing (SU) has been widely investigated for hyperspectral analysis with the aim to find the optimal subset of spectral signatures in a spectral library (known in advance) that can optimally model each pixel of the given hyperspectral image. Usually, the available spectral library organizes spectral signatures in groups. However, most existing strategies do not take full advantage of the inherent properties in the library. In this article, we design a convex framework for SU that incorporates the group structure of the spectral library. The convex framework includes two kinds of algorithms derived from either the primal or the dual form of the alternating direction method of multipliers (ADMM). Then, the convergence properties of the convex framework are established. Based on the convex framework, a novel nonconvex framework is developed for unmixing, which provides a new manner to enhance the sparsity of solution. The core of the nonconvex framework is to design a nonconvex penalty function for efficient minimization utilizing the generalized shrinkage mapping. The penalty function can be regarded as a closer approximation of the$l_{0}$norm. Experiments conducted on simulated and real hyperspectral data demonstrate the superiority and effectiveness of the proposed nonconvex framework in improving the unmixing performance and enhancing the sparsity of solution with respect to state-of-the-art techniques.

Passively sounding icy and rocky bodies in our solar system provides a way to observe the surface and subsurface of these objects without the need for costly transmitters. Jupiter’s decametric radiation provides a suitable source of radio frequency signals for sounding on geological scales of interest, but its spectral structure can introduce undesired artifacts. Recent studies have shown that Jovian bursts could be effectively applied as a source of passive sounding. However, it has been noted that the spectral properties of Jovian bursts, as measured with JUNO data, are not perfectly modeled as white noise, which can result in degraded sensitivity and vertical resolution for passive sounding. In response to this finding, we present conditioning processes that improve the echo detectability and sounding resolution for Jovian burst-like signals. More than 18 h of Jovian burst recordings are used to simulate conditioning of the natural spectral variation, demonstrating that high-quality corrections are possible with these processes for noise conditions in both the Jovian and Earth/Moon systems.

Radar sounders (RS) are gaining importance in planetary missions thanks to their unique capability of providing direct measurements of subsurface structures. To support their design and data interpretation, several electromagnetic (e.m.) simulation techniques have been developed with enhanced capabilities of emulating the RS acquisition process. However, the raw simulated radargrams obtained from e.m. simulators are difficult to interpret and analyze without a focusing operation, which results in an underestimation of the RS detection performance. While frequency methods for range and azimuth compression of real RS data are well-established, their use on simulated data is not addressed in the literature and requires major modifications. This paper presents a novel method that implements azimuth compression using unfocused and focused processing on simulated raw data. The proposed method is based on an adaptation of the Range-Doppler algorithm to the case of raw data generated by a coherent RS simulator. The method is demonstrated in three case-studies to show the similarity between simulated and real data processing: 1) simple geometries; 2) a simulated SHARAD radargram compared with the real data product; and 3) a real application scenario for supporting the design of a new RS instrument. The results indicate higher fidelity of the focused simulated data with the real data product and the target structure, confirming the usefulness of the proposed approach in obtaining a realistic processing of simulated radargrams.

A satellite mission onboard a radar sounder for the observation of the earth’s polar regions can greatly support the monitoring of the cryosphere and climate change analyses. Several studies are in progress proposing the design and demonstrating the performance of such an earth-orbiting radar sounder (EORS). However, one critical aspect of the cryospheric targets that are often ignored and simplified in these studies is the complex geoelectrical nature of the polar ice. In this article, we present a performance assessment of the polar ice target detectability by focusing on their realistic representation. This is obtained by simulating the orbital radargrams corresponding to different regions of the polar cryosphere by leveraging the data available from airborne campaigns in Antarctica and Greenland. We propose novel performance metrics to analyze the detectability of the internal reflecting horizons (IRHs), the basal interface, and to analyze the nature of the basal interface. This performance assessment strategy can be applied to guide the design of the signal-to-noise ratio (SNR) budget at the surface, which can further support the selection of the main orbital instrument parameters, such as the transmitted power, the two-way antenna gain, and the processing gains.

Deep learning-based methods (e.g., convolutional neural network (CNN)-based methods), have shown increasing potential in hyperspectral image (HSI) change detection (CD). However, the recent advances in CNN-based methods in HSI CD tasks are mostly devoted to designing more complex architectures or adding additional hand-designed blocks. This increases the number of parameters making model training difficult. In this paper, we propose an end-to-end residual self-calibrated network (RSCNet) to increase the accuracy of HSI CD. To fully exploit the spatial information, the proposed RSCNet method adaptively builds inter-spatial and inter-spectral dependencies around each spatial location with fewer extra parameters and reduced complexity. Moreover, the introduced self-calibrated convolution (SCConv) helps to generate more discriminative representations by heterogeneously exploiting convolutional filters nested in the convolutional layer. The designed RSC module can explicitly incorporate richer information by introducing response calibration operation. The experiments on four bi-temporal HSI datasets demonstrated that the proposed RSCNet method is more accurate than ten widely used benchmark methods.

The very high spectral resolution in hyperspectral images (HSIs) offers an opportunity to detect subtle land-cover changes. However, the availability of HSIs acquired from different platforms requires the development of change detection (CD) methods capable of processing HSIs with different spatial resolutions. In this paper, we propose a general end-to-end subpixel convolution-based residual network (SPCNet) to accomplish the CD task between high spatial resolution (HR) and low spatial resolution (LR) HSIs. To effectively tackle the resolution matching issue, a super resolution (SR) block with an efficient subpixel convolution layer is introduced to upscale the LR feature maps into HR maps. The subpixel convolution layer can fully explore the subpixel context information by learning an array of upscaling filters. Moreover, the designed SPC module is embedded into the LR branch to generate more discriminative representations. More importantly, the SPC module as a plug-and-play unit has the potential to be embedded into other baseline networks to enhance the feature learning capability. Experimental results on four HSI datasets demonstrate the effectiveness of the proposed SPCNet.

To infer unknown remote sensing scenarios, most existing technologies use a supervised learning paradigm to train deep neural network (DNN) models on closed datasets. This paradigm faces challenges such as highly spatiotemporal variants and ever-changing scale-heterogeneous remote sensing scenarios. Additionally, DNN models cannot scale to new scenarios. Lifelong learning is an effective solution to these problems. Current lifelong learning approaches focus on overcoming thecatastrophic forgettingissue (i.e., a successive increase in heterogeneous remote sensing scenes causes models to forget historical scenes) while ignoring theknowledge recallissue (i.e., how to facilitate the learning of new scenes by recalling historical experiences), which is a significant problem. This paper proposes a lifelong learning framework called asymmetric collaborative network (SCN) for lifelong remote sensing image classification. This framework consists of two structurally distinct networks: a preserving network (Pres-Net) and a transient network (Trans-Net), which imitates the long- and short-term memory processes in the brain, respectively. Moreover, this framework is based on two synergistic knowledge transfer mechanisms: triple distillation and prior feature fusion. The triple distillation mechanism enables knowledge persistence from Trans-Net to Pres-Net to achieve better memorization; the prior feature fusion mechanism enables knowledge transfer from Pres-Net to Trans-Net to achieve better recall. Experiments on three open datasets demonstrate the effectiveness of SCN for 3-, 6-, and 9-task-length learning. The idea of asymmetric separation networks and the synergistic strategy proposed in this paper are expected to provide new solutions to the translatability of the classification of remote sensing images in real world scenarios.

The current remote sensing (RS) open data policy for multispectral (MS) missions such as Sentinel-2 and Landsat-8, together with the availability of free cloud distributed processing platforms such as Google Earth Engine, makes it possible the quick generation of burned area (BA) products even for nonexperts in the field. Indeed, fires and BAs can be detected using burn severity indices, which are usually obtained by simple band algebra operations. However, simple approaches can aid BA estimation only if typical error patterns are known and accounted for, especially when working at large (e.g., continental) scales. This article proposes an automatic BA detection system based on burn severity index thresholding, which integrates dedicated false and missed alarm mitigation strategies to improve the detection accuracy. The system is tested on Sentinel-2 and Landsat-8 data over ten different locations in Europe and spanning year 2018. Three known burn severity indices plus a custom one defined to improve the performance in the considered study area are under study. Results show that burned index thresholding is possible within accuracy bounds slightly larger than the state of the art, which is acceptable by considering the proposed simplified processing framework.

Among various multimodal remote sensing data, the pairing of multispectral (MS) and panchromatic (PAN) images is widely used in remote sensing applications. This article proposes a novel global collaborative fusion network (GCFnet) for joint classification of MS and PAN images. In particular, a global patch-free classification scheme based on an encoder-decoder deep learning (DL) network is developed to exploit context dependencies in the image. The proposed GCFnet is designed based on a novel collaborative fusion architecture, which mainly contains three parts: 1) two shallow-to-deep feature fusion branches related to individual MS and PAN images; 2) a multiscale cross-modal feature fusion branch of the two images, where an adaptive loss weighted fusion strategy is designed to calculate the total loss of two individual and the cross-modal branches; 3) a probability weighted decision fusion strategy for the fusion of the classification results of three branches to further improve the classification performance. Experimental results obtained on three real datasets covering complex urban scenarios confirm the effectiveness of the proposed GCFnet in terms of higher accuracy and robustness compared to existing methods. By utilizing both sampled and non-sampled position data in the feature extraction process, the proposed GCFnet can achieve excellent performance even in a small sample-size case. The codes will be available from the website: https://github.com/SicongLiuRS/GCFnet.

Due to the scarcity of labeled samples, clustering in hyperspectral images (HSIs) has a great potential and application value. However, current clustering methods are mainly pixel-level techniques that neglect the large spectral variability of a scene and suffer from massive time and memory consumption when dealing with large HSIs. In this article, we propose a superpixel-level global and local similarity graph-based clustering (SGLSC) algorithm that can classify ground objects exploiting spectral and spatial dimensions with reasonable time and memory consumption on large HSIs. The proposed SGLSC exploits the superpixel concept, which is treated as a homogeneous entity, into the clustering process. For modeling the essential structure of HSIs, a similarity graph combing the global and local information is constructed and inserted into the spectral clustering to partition the superpixel-level graph structure. The proposed method was tested on three benchmark HSIs’ datasets and compared with some advanced literature algorithms. Experiments demonstrate that it can obtain promising results.

Building extraction in VHR RSIs remains a challenging task due to occlusion and boundary ambiguity problems. Although conventional convolutional neural networks (CNNs) based methods are capable of exploiting local texture and context information, they fail to capture the shape patterns of buildings, which is a necessary constraint in the human recognition. To address this issue, we propose an adversarial shape learning network (ASLNet) to model the building shape patterns that improve the accuracy of building segmentation. In the proposed ASLNet, we introduce the adversarial learning strategy to explicitly model the shape constraints, as well as a CNN shape regularizer to strengthen the embedding of shape features. To assess the geometric accuracy of building segmentation results, we introduced several object-based quality assessment metrics. Experiments on two open benchmark datasets show that the proposed ASLNet improves both the pixel-based accuracy and the object-based quality measurements by a large margin. The code is available at: https://github.com/ggsDing/ASLNet.

Deep Learning (DL) methods are widely used for Change Detection (CD) in multi-temporal Remote Sensing (RS) images. The recently reported unsupervised DL CD methods alleviate the problem of the labeled data collection affecting the supervised ones. Many of them exploit the DL models (e.g., Convolutional Autoencoder (CAE)) as a feature extractor and use the retrieved features to detect the changes. However, these features do not efficiently preserve the geometrical details, and they do not optimize the selection of informative features for change detection. We propose an unsupervised DL CD method that exploits the features extracted by a CAE trained with a super-resolution based loss function. The loss function allows the CAE to be trained to reconstruct the spatial information thus generating features preserving the geometrical details. The proposed method exploits a feature selection based on the Structured Similarity Index (SSIM) to perform a texture analysis and chooses couples of bi-temporal features providing relevant information about changes. We tested the proposed method on a couple of bi-temporal Landsat-8 images representing a burned area near Granada, Spain.

This paper presents the SaTellite RAdar sounder for earTh sUbsurface Sensing (STRATUS), which is a satellite mission for Earth Observation (EO) with an onboard instrument capable of probing the Earth's subsurface in polar and arid regions. STRATUS is based on an innovative distributed radar sounder (RS) with the unique capability to obtain continuous and large-scale subsurface measurements, with homogeneous and consistent quality in two of the least characterized and crucial frontiers of Earth: globally on the polar ice sheets, i.e., Greenland and Antarctica (primary objective), and regionally on the arid areas and deserts. STRATUS is a ground-breaking exploratory mission addressing crucial scientific questions. It provides new fundamental data that have not been acquired by any other past or present remote sensing mission on the Earth, with an expected high and genuine scientific return enabling the assessment of the climate change signature in the Earth subsurface.

The availability of multi-sensor data presents an opportunity for change detection based on the complementary use of properties associated with different data sources. This paper proposes a new unsupervised change detection framework based on the joint use of SAR and optical images. The framework exploits a contrastive learning algorithm and the assumption of the scarcity of relevant changes. The proposed architecture is a pseudo-Siamese network, which is trained to regress the feature vector of bi-temporal concatenated SAR-optical input data. The output feature vectors from two branches of the pseudo-Siamese network are used to calculate change intensity maps. Then, the binary change map is obtained by setting a proper threshold. The proposed method is validated by using a multi-sensor dataset made up of Sentinel-1 and Sentinel-2 images that is also compared with the single use of each modality image. Experimental results demonstrate improvements of the multisensor approach over single modality and confirm the potentiality of the joint use of SAR and optical images in change detection.

Radar sounder data are widely used for investigating geological structures and processes in the subsurface of icy and arid areas. Visual interpretation is one of the main techniques used in the literature to extract information from radargrams. There exist some automatic approaches but mostly supervised. However, no methods exploit deep learning in an unsupervised way. Here, we propose an automatic and unsupervised technique for extracting information on the subsurface geological targets. The technique is built upon three steps: i) generation of a coarse segmentation map based on the radargram statistical properties, ii) refinement of the coarse map with deep learning to detect target reflections, and iii) analysis of the deep features to identify buried targets. We tested the proposed method on MARSIS radar data acquired near the South Pole of Mars. The experimental results prove the effectiveness of the proposed method.

We describe a CubeSat constellation integrated with ultrawideband (UWB) multiple input and multiple output (MIMO) radars operating over the frequency range of 50–600 MHz for remote sensing of large ice sheets on the Earth, as well as distant planets and moons. It can be used for measurements of ice thickness and basal conditions of Antarctica and Greenland ice sheets, the mapping of lunar lava tubes, and sub-surface detection/characterization of volatiles on Mars. In this paper we provide a short discussion of the science requirements and conceptual design of a constellation of CubeSats with UWB MIMO radars for sounding ice and mapping of lava tubes.

Recent satellite missions have initiated a new era in the area of Satellite Image Time Series (SITS) analysis by providing a huge number of High Resolution (HR) spectral-temporal images. The availability of HR images opens a door to an unprecedented wide range of possibilities to produce and develop high resolution Land Cover (LC) and Land Cover Change (LCC) maps. The goal of this paper is to effectively use high spatio-temporal resolution images to generate LCC maps by defining a novel automatic and unsupervised deep learning method based on three-dimensional (3D) Convolutional Neural Network (CNN). The method extracts spatio-temporal information from long SITS by using a pre-trained 3D CNN, detects changes and locates them in space and time. Experiments have provided promising results over both Amazonia and Saudi Arabia in the period 2013–2017, and has been compared to the other well-known LCC detection method.

The prediction of forest biophysical parameters is an important task in remote sensing for understanding global carbon cycle. Spectral remote sensing data are available globally at a relatively economical cost making them a viable resource for forest remote sensing. However, the main drawbacks associated with such data is the uncertainty of predictions and cluttered process of selecting band combinations from hyperspectral/multispectral data to produce spectral features for modelling. In this paper, we present an approach that exploits the latest developments in generative variational autoencoders (VAE) that produce disentangled representation from input data to assess the capability of hyperspectral data to model forest aboveground biomass (AGB). The proposed VAE generates a special kind of deep spectral features that are proportional to AGB. A modelling accuracy of R2 = 0.57 (cross-validated) was obtained by the proposed approach, thus pointing out the potential of hyperspectral data to model AGB using disentangled deep spectral features. The proposed approach also enables in bypassing the unreliable process of selecting band combinations to produce spectral features and shows good prospects for mapping global level biomass.

The accurate classification of remote sensing (RS) data at large scale is typically hampered by the availability of training data representative of the whole study area. To solve this problem, we propose a method that aims to enlarge existing training sets leveraging publicly available thematic products. First, the available thematic product of the target domain ($D_{T}$) (RS data geographically distant from the training samples) is processed to extract few labeled target samples. These labeled target samples are jointly used with the annotated samples of the source domain ($D_{S}$) (RS data where training set is available) to find a mapping space where the data are aligned. This common latent space allows us to enlarge the training set in an unsupervised (no annotated samples from the$D_{T}$are required) but reliable way. The results obtained in Amazon using the Copernicus Global Land Service - Land cover (CGLS-LC) map demonstrate the effectiveness of the method. The enlarged training set achieves an Overall Accuracy (OA) of 87% compared to 80% obtained with the initial training set.

The monitoring of the snow cover area (SCA) from optical sensors on board of satellites is affected by the trade-off between spatial and temporal resolution provided by the current operational missions. This limits the possibility to exploit satellite SCA for hydrological purposes. In this paper, we propose a novel downscaling approach driven by the low resolution (LR) information that takes advantage of i) all the high resolution (HR) images acquired in the past over a catchment; and ii) the geomorphometric features that drive the snow redistribution process. Possible applications of the proposed method are time-series gap-filling and snow pattern detection. The downscaled scenes have been validated across existing HR scenes showing an accuracy of about 90%.

The rapid and explosive growth of remote sensing image dataset (e.g., optical, SAR, LiDAR) promotes the development of the aerospace industry. However, images with complex coverage scenes are usually captured by either different sensors from different perspectives or the same sensor in different periods [1]. These factors have brought a great challenge to precision image co-registration, and it is difficult to identify a fully universal method to cope with all registration cases. Any kind of image registration algorithm needs to consider the imaging principle, radiometric and geometric distortions, noise interference, and so on. To date, numerous efforts have been made to overcome these challenges and improve the performance of multimodal remote sensing image registration, which can be classified into three categories: area-based methods, feature-based methods and a joint of previous two categories [2].

To overcome the limited capability of most state-of-the-art change detection (CD) methods in modeling spatial context of multispectral high spatial resolution (HR) images and exploiting all spectral bands jointly, this letter presents a novel unsupervised deep-learning-based CD method that can effectively model contextual information and handle the large number of bands in multispectral HR images. This is achieved by exploiting all spectral bands after grouping them into spectral-dedicated band groups. To eliminate the necessity of multitemporal training data, the proposed method exploits a data set targeted for image classification to train spectral-dedicated Auxiliary Classifier Generative Adversarial Networks (ACGANs). They are used to obtain pixelwise deep change hypervector from multitemporal images. Each feature in deep change hypervector is analyzed based on the magnitude to identify changed pixels. An ensemble decision fusion strategy is used to combine change information from different features. Experimental results on the urban, Alpine, and agricultural Sentinel-2 data sets confirm the effectiveness of the proposed method.

Most change detection (CD) methods are unsupervised as collecting substantial multitemporal training data is challenging. Unsupervised CD methods are driven by heuristics and lack the capability to learn from data. However, in many real-world applications, it is possible to collect a small amount of labeled data scattered across the analyzed scene. Such a few scattered labeled samples in the pool of unlabeled samples can be effectively handled by graph convolutional network (GCN) that has recently shown good performance in semisupervised single-date analysis, to improve change detection performance. Based on this, we propose a semisupervised CD method that encodes multitemporal images as a graph via multiscale parcel segmentation that effectively captures the spatial and spectral aspects of the multitemporal images. The graph is further processed through GCN to learn a multitemporal model. Information from the labeled parcels is propagated to the unlabeled ones over training iterations. By exploiting the homogeneity of the parcels, the model is used to infer the label at a pixel level. To show the effectiveness of the proposed method, we tested it on a multitemporal Very High spatial Resolution (VHR) data set acquired by Pleiades sensor over Trento, Italy.

Convolutional neural networks (CNNs) are widely used in the field of remote sensing images. However, the applications of CNNs and related techniques often ignore the properties of remote sensing data. In our study, we focus on the hyperspectral image (HSI) classification problem, and address the issue of including the very rich spectral information present in HSIs in CNN-based models to produce highly accurate classification results. We propose a two-step classification technique, ClusterCNN. The first step divides HSI pixels into different clusters, to form a material map which can be considered as a compressed expression of the original spectral features. The second step trains a CNN that can extract spatial features from the material map, and then exploits these spatial features to classify HSI pixels. The proposed approach follows a strict hierarchy to exploit both the spectral and spatial features in HSIs. Experimental results show the effectiveness of ClusterCNN as compared to the much more complicated state-of-the-art approaches.

To reduce the storage requirements, remote-sensing (RS) images are usually stored in compressed format. Existing scene classification approaches using deep neural networks (DNNs) require to fully decompress the images, which is a computationally demanding task in operational applications. To address this issue, in this article, we propose a novel approach to achieve scene classification in Joint Photographic Experts Group (JPEG) 2000 compressed RS images. The proposed approach consists of two main steps: 1) approximation of the finer resolution subbands of reversible biorthogonal wavelet filters used in JPEG 2000 and 2) characterization of the high-level semantic content of approximated wavelet subbands and scene classification based on the learned descriptors. This is achieved by taking codestreams associated with the coarsest resolution wavelet subband as input to approximate finer resolution subbands using a number of transposed convolutional layers. Then, a series of convolutional layers models the high-level semantic content of the approximated wavelet subband. Thus, the proposed approach models the multiresolution paradigm given in the JPEG 2000 compression algorithm in an end-to-end trainable unified neural network. In the classification stage, the proposed approach takes only the coarsest resolution wavelet subbands as input, thereby reducing the time required to apply decoding. Experimental results performed on two benchmark aerial image archives demonstrate that the proposed approach significantly reduces the computational time with similar classification accuracies when compared with traditional RS scene classification approaches (which requires full image decompression).

Recent studies have proposed that Jovian decametric radiation (DAM) can be effectively exploited for probing the subsurface of Jupiter's icy moons by using passive radio sounding. However, these studies were based on the assumption that Jovian noise is white and stationary. Therefore, additional investigations into the temporal stability, predictability, and spectral properties of Jovian noise are required to fully assess the potential of passive radar sounding and improve the acquisitions planning. In this article, we investigate these properties of the Jovian DAM to understand their impact on radar sounding performance. This is done by analyzing the recently available radio spectra acquired by the JUNO Waves instrument. Results are also evaluated for the specific case of the Radar for Icy Moon Exploration (RIME) and the Radar for Europa Assessment and Sounding: Ocean to Near-surface (REASON) that have been selected for European Space Agency (ESA) and NASA missions to Ganymede and Europa. Our results show that the Jovian DAM is not perfectly white, but no severe distortions in the range response should be expected. The results on spatiotemporal occurrence show that Jupiter's DAM has a variable probability of occurrence, which is rather sporadic for some frequency ranges. The results on RIME and REASON flybys show that the Jovian DAM occurrence probability is relatively low for selected sub-Jovian flybys at Europa. For the RIME Ganymede orbital phase, a large number of high occurrence passive acquisition opportunities are expected. The experimental results also show that a large bandwidth receiver would enhance the chance of recording Jovian noise.

Radar sounder surface imaging is a rather unexplored approach to the analysis of planetary bodies. While a radar sounder is an instrument specifically designed for subsurface investigations, a particular set of power measurements (denoted as off-nadir surface echoes) can be exploited together with an external digital elevation model to produce images of the investigated surface at meters wavelength. The use of the off-nadir data may also reveal the presence of previously undetected subsurface features. In this article, we present a method for producing surface roughness images by high-frequency (HF) radar sounder data. The study of surface roughness in the HF band is particularly useful for both geologic studies and landing-zone reconnaissance as it is evaluated at meters to hundreds of meters horizontal scale. The proposed method combines off-nadir data of the Shallow Radar Sounder (SHARAD) with the Mars Orbiter Laser Altimeter (MOLA) digital elevation model. The produced roughness images at 20 MHz (15-m wavelength) of the Martian surface provide higher coverage and resolution of the surface roughness characterization at a 10-100-m horizontal scale than previous SHARAD work. By comparing the experimental roughness image with the one obtained by radar backscattering simulations, it is possible to identify subsurface features. In our experiments, we were able to produce a bidimensional image of a previously undetected large buried crater (10 km ×12 km) located in the Nili Fossae. This finding opens up new possibilities in exploiting radar sounding data for better detecting shallow subsurface features.

Atmospheric phase screen (APS) is a very critical issue for the application of interferometric synthetic aperture radar (InSAR) techniques. The spatial-temporal variations of APS are the dominant error source in interferograms and may completely mask displacement signals. Many external meteorological data-based methods and phase-based methods have been developed in the past decades, but all have their inherent limitations. In this article, we propose a deep learning-based method, which is based on an attention-based deep residual U-shaped network (ARU-Net), to mitigate atmospheric artifacts. With this approach, APS patches and clean interferogram patches are sampled from InSAR interferograms to train the network. After training, the network can be used to mitigate the APS for individual interferograms. Compared with the generic atmospheric correction model (GACOS) and the advanced time-series InSAR method distributed scatterer interferometry (DSI), the key advantage of our method is that atmospheric delay can be effectively learned and removed from individual high-resolution interferometric phase itself without external data. Accuracy was validated by using individual and stacked interferograms from TerraSAR-X data over the Hong Kong International Airport (HKIA) and Hong Kong Science Park (HKSP) sites. The results showed that our method consistently delivered greater standard deviation (SD) reduction after APS correction than the GACOS method. Moreover, the time-series results were in agreement with the DSI and leveling measurements. The effectiveness of the proposed ARU-Net to remove APS effects from interferograms shows great potential for the development of a new set of deep learning-based APS reduction methods.

The binary segmentation of roads in very high resolution (VHR) remote sensing images (RSIs) has always been a challenging task due to factors such as occlusions (caused by shadows, trees, buildings, etc.) and the intraclass variances of road surfaces. The wide use of convolutional neural networks (CNNs) has greatly improved the segmentation accuracy and made the task end-to-end trainable. However, there are still margins to improve in terms of the completeness and connectivity of the results. In this article, we consider the specific context of road extraction and present a direction-aware residual network (DiResNet) that includes three main contributions: 1) an asymmetric residual segmentation network with deconvolutional layers and a structural supervision to enhance the learning of road topology (DiResSeg); 2) a pixel-level supervision of local directions to enhance the embedding of linear features; and 3) a refinement network to optimize the segmentation results (DiResRef). Ablation studies on two benchmark data sets (the Massachusetts data set and the DeepGlobe data set) have confirmed the effectiveness of the presented designs. Comparative experiments with other approaches show that the proposed method has advantages in both overall accuracy and F1-score. The code is available at:https://github.com/ggsDing/DiResNet.

The trade-off between feature representation power and spatial localization accuracy is crucial for the dense classification/semantic segmentation of remote sensing images (RSIs). High-level features extracted from the late layers of a neural network are rich in semantic information, yet have blurred spatial details; low-level features extracted from the early layers of a network contain more pixel-level information but are isolated and noisy. It is therefore difficult to bridge the gap between high- and low-level features due to their difference in terms of physical information content and spatial distribution. In this article, we contribute to solve this problem by enhancing the feature representation in two ways. On the one hand, a patch attention module (PAM) is proposed to enhance the embedding of context information based on a patchwise calculation of local attention. On the other hand, an attention embedding module (AEM) is proposed to enrich the semantic information of low-level features by embedding local focus from high-level features. Both proposed modules are lightweight and can be applied to process the extracted features of convolutional neural networks (CNNs). Experiments show that, by integrating the proposed modules into a baseline fully convolutional network (FCN), the resulting local attention network (LANet) greatly improves the performance over the baseline and outperforms other attention-based methods on two RSI data sets.

Crown features derived from high-density airborne laser scanning (ALS) data have proven to be effective for forest species classification at the individual tree level. Most of the general state-of-the-art (SoA) techniques rely on coarse-level crown features extracted from ALS data and under-utilize both the spatial and the spectral information available in the point clouds, Moreover, they are designed on the expected properties of the specific analyzed forest. We present a novel species classification approach, based on quantization of the entire 3-D tree crown into smaller elementary crown volumes (ECVs) that effectively captures the spatial distribution of filled (i.e., stem, branch, and foliage) and empty volumes of crowns. In the first step, a data-driven process dynamically tests and compares three quantization strategies to tailor the definition of the ECV to the forest type (e.g., conifer and deciduous forest). In the second step, for each ECV, a histogram vector is made up of features representing the light detection and ranging (LiDAR) point distribution and intensity to model the internal and the external local crown characteristics. Then, tree histogram feature vectors are obtained by stacking all the ECV histogram feature vectors. Finally, classification is performed by a support vector machine (SVM) classifier using the histogram intersection kernel. All experiments were performed on three high-density (50-200 points/m2) ALS data sets of deciduous, conifer, and mixed (i.e., both deciduous and conifer) trees. The higher classification accuracy of the proposed method over the SoA one proves its ability to better capture the crown characteristics of individual trees, including species-specific traits.

Supervised classification algorithms require a sufficiently large set of representative training samples to generate accurate land-cover maps. Collecting reference data is difficult, expensive, and unfeasible at the large scale. To solve this problem, this article introduces a novel approach that aims to extract reliable labeled data from existing thematic products. Although these products represent a potentially useful information source, their use is not straightforward. They are not completely reliable since they may present classification errors. They are typically aggregated at polygon level, where polygons do not necessarily correspond to homogeneous areas. Finally, usually, there is a semantic gap between map legends and remote sensing (RS) data. In this context, we propose an approach that aims to: 1) perform a domain understanding to detect the discrepancies between the thematic map domain and the RS data domain; 2) use RS data contemporary to the map to decompose the thematic product from the semantic and spatial viewpoints; and 3) extract a database of informative and reliable training samples. The database of weak labeled units is used for training an ensemble of classifiers on recent data whose results are then combined in a majority voting rule. Two sets of experimental results obtained on MS images by extracting training samples from a crop type map and the 2018 Corine Land Cover (CLC) map, respectively, confirm the effectiveness of the proposed approach.

Change detection (CD) is one of the main applications of remote sensing. With the increasing popularity of deep learning, most recent developments of CD methods have introduced the use of deep learning techniques to increase the accuracy and automation level over traditional methods. However, when using supervised CD methods, a large amount of labeled data is needed to train deep convolutional networks with millions of parameters. These labeled data are difficult to acquire for CD tasks. To address this limitation, a novel semisupervised convolutional network for CD (SemiCDNet) is proposed based on a generative adversarial network (GAN). First, both the labeled data and unlabeled data are input into the segmentation network to produce initial predictions and entropy maps. Then, to exploit the potential of unlabeled data, two discriminators are adopted to enforce the feature distribution consistency of segmentation maps and entropy maps between the labeled and unlabeled data. During the competitive training, the generator is continuously regularized by utilizing the unlabeled information, thus improving its generalization capability. The effectiveness and reliability of our proposed method are verified on two high-resolution remote sensing data sets. Extensive experimental results demonstrate the superiority of the proposed method against other state-of-the-art approaches.

Hyperspectral images (HSIs) provide rich spectral-spatial information with stacked hundreds of contiguous narrowbands. Due to the existence of noise and band correlation, the selection of informative spectral-spatial kernel features poses a challenge. This is often addressed by using convolutional neural networks (CNNs) with receptive field (RF) having fixed sizes. However, these solutions cannot enable neurons to effectively adjust RF sizes and cross-channel dependencies when forward and backward propagations are used to optimize the network. In this article, we present an attention-based adaptive spectral-spatial kernel improved residual network (A2S2K-ResNet) with spectral attention to capture discriminative spectral-spatial features for HSI classification in an end-to-end training fashion. In particular, the proposed network learns selective 3-D convolutional kernels to jointly extract spectral-spatial features using improved 3-D ResBlocks and adopts an efficient feature recalibration (EFR) mechanism to boost the classification performance. Extensive experiments are performed on three well-known hyperspectral data sets, i.e., IP, KSC, and UP, and the proposed A2S2K-ResNet can provide better classification results in terms of overall accuracy (OA), average accuracy (AA), and Kappa compared with the existing methods investigated. The source code will be made available at https://github.com/suvojit- 0×55aa/A2S2K-ResNet.

Building change detection (CD), important for its application in urban monitoring, can be performed in near real time by comparing prechange and postchange very-high-spatial-resolution (VHR) synthetic-aperture-radar (SAR) images. However, multitemporal VHR SAR images are complex as they show high spatial correlation, prone to shadows, and show an inhomogeneous signature. Spatial context needs to be taken into account to effectively detect a change in such images. Recently, convolutional-neural-network (CNN)-based transfer learning techniques have shown strong performance for CD in VHR multispectral images. However, its direct use for SAR CD is impeded by the absence of labeled SAR data and, thus, pretrained networks. To overcome this, we exploit the availability of paired unlabeled SAR and optical images to train for the suboptimal task of transcoding SAR images into optical images using a cycle-consistent generative adversarial network (CycleGAN). The CycleGAN consists of two generator networks: one for transcoding SAR images into the optical image domain and the other for projecting optical images into the SAR image domain. After unsupervised training, the generator transcoding SAR images into optical ones is used as a bitemporal deep feature extractor to extract optical-like features from bitemporal SAR images. Thus, deep change vector analysis (DCVA) and fuzzy rules can be applied to identify changed buildings (new/destroyed). We validate our method on two data sets made up of pairs of bitemporal VHR SAR images on the city of L'Aquila (Italy) and Trento (Italy).

Nowadays, deep learning (DL) finds application in a large number of scientific fields, among which the estimation and the enhancement of signals disrupted by the noise of different natures. In this article, we address the problem of the estimation of the interferometric parameters from synthetic aperture radar (SAR) data. In particular, we combine convolutional neural networks together with the concept of residual learning to define a novel architecture, named Φ-Net, for the joint estimation of the interferometric phase and coherence. Φ-Net is trained using synthetic data obtained by an innovative strategy based on the theoretical modeling of the physics behind the SAR acquisition principle. This strategy allows the network to generalize the estimation problem with respect to: 1) different noise levels; 2) the nature of the imaged target on the ground; and 3) the acquisition geometry. We then analyze the Φ-Net performance on an independent data set of synthesized interferometric data, as well as on real InSAR data from the TanDEM-X and Sentinel-1 missions. The proposed architecture provides better results with respect to state-of-the-art InSAR algorithms on both synthetic and real test data. Finally, we perform an application-oriented study on the retrieval of the topographic information, which shows that Φ-Net is a strong candidate for the generation of high-quality digital elevation models at a resolution close to the one of the original single-look complex data.

Radar sounders (RSs) are low-frequency instruments that profile the shallow subsurface of planetary targets to obtain valuable scientific information. The prediction of the RS performance and the interpretation of the target properties from the RS data are challenging due to the complex electromagnetic interaction among many acquisition variables. Simulation of RS data can address this issue by modeling the complex interaction and producing simulated radargrams representing the acquisition scenario. In this article, we present an approach to generate databases of geoelectrical models and simulated radargrams corresponding to combinations of: 1) target geoelectrical hypotheses; 2) RS parameters; and 3) acquisition geometry configurations. The proposed approach exploits this database for: 1) predicting the detection performance and sensitivity of the RS and 2) understanding the interpretability of the underlying hypotheses. In order to identify hypothesis combinations that can be unambiguously inverted from the radargrams, we analyze the similarity between pairs of geoelectrical models and between the simulated radargrams, and the statistical distance between radargram features. The approach is demonstrated for the case of Radar for Icy Moons Exploration (RIME), using three selected targets on the Jovian moon Ganymede, with three different simulation techniques. The results are very promising and reveal the effectiveness of the proposed approach in extracting valuable information regarding: 1) the target detection performance of RIME; 2) the sensitivity to the dielectric contrast; 3) the separability of radargram features; and 4) the identification of hypothesis combinations producing significantly different radar response, and thus invertible.

Band selection (BS) can mitigate the “curse of dimensionality” problem and improve the performance of hyperspectral image (HSI) classification. Genetic algorithms (GAs) have been applied to the task of hyperspectral BS showing significant advantages compared with other literature methods. However, the traditional GAs-based methods often select sets of bands having residual redundancy due to the large search space related to hyperspectral BS and the limitation of premature convergence in GAs. Moreover, existing GAs-based methods often are supervised, and that needs a large number of labeled samples to compute the fitness value for assessing the quality of selected bands. In this article, an unsupervised BS approach based on an improved GA is proposed. A fitness function based on the fisher score combined with superpixel is designed for evaluating the discriminability of band subsets considering both spectral and spatial information. Then, modified genetic operations are constructed to restrain the search space and reduce the redundancy of selected bands. The performance of the proposed spectral-spatial GA-based BS method is evaluated on three HSIs. The experimental results demonstrate that the proposed method is superior to the traditional GA-based method and seven state-of-the-art unsupervised methods.

This paper presents the Subsurface Radar Sounder (SRS) instrument onboard European Space Agency's (ESA) EnVision mission. EnVision is one of the three candidates selected for the Cosmic Vision 2015-2025 M5 medium-class missions. It is aimed at exploring the activity, the geologic history and the atmosphere of Venus. SRS is an orbital ground-penetrating radar with the unique science objectives of understanding the evolution of Venus' surface by searching for subsurface dielectric interfaces in the top hundreds of metres of the crust. In the paper, we describe the main science objectives of SRS, the performance evaluation under expected target conditions, the instrument design and the acquisition strategy that maximize the scientific returns.

This paper presents an approach for large-scale precise mapping of agricultural fields based on the analysis of Satellite Image Time Series (SITS) acquired by ESA Sentinel-2 (S2) satellite constellation. The approach has been developed in the framework of the ESA SEOM - Scientific Exploitation of Operational Missions - S2-4Sci Land and Water project. The goal is to design a flexible and automatic processing chain able to perform mapping in massive data. Here we focus on precision agriculture products generation at country level. In particular, the Country of study is Italy and the application goal is precision agriculture of single crop fields. To achieve this goal, two macro challenges are considered: (i) download and pre-processing of S2 SITS, and (ii) multi-temporal (MT) fine characterization of agricultural fields. Both challenges are addressed in an automatic way by exploiting and/or updating state-of-the-art methodologies. Promising results have been obtained over years 2017 and 2018 for Italy.

Data acquistions performed by instruments in the scientific payload of planetary missions require an accurate planning, for obtaining the expected science return. This paper presents an approach to the automatic planning and scheduling of acquisitions of remote sensing instruments in planetary missions. The approach exploits multi-objective optimization techniques based on Genetic Algorithms (GAs) to produce schedules that are optimized tradeoffs of multiple criteria identifying the quality of the considered acquisition campaign. The approach is general, but in this paper we present it considering radar sounder instruments. Accordingly, we design a few ad-hoc quality metrics for radar sounder (RS) observations and apply the proposed approach to the operations of RIME (Radar for Icy Moons Exploration), which is the radar sounder designed for the JUpiter ICy moons Explorer (JUICE) mission.

Very-high-resolution (VHR) multi-temporal images are important in remote sensing to monitor the dynamics of the Earth surface. Image semantic segmentation classifies pixels and assigns them label from meaningful object groups. It has been extensively studied in context of single image analysis, however not explored for multi-temporal one. In this paper we propose to extend supervised semantic segmentation to the unsupervised joint segmentation of multi-temporal images. The proposed method processes multi-temporal images by separately feeding them to a deep network comprising of trainable convolutional layers. The training process does not involve any external label. Segmentation labels are obtained from argmax classification of the final layer. Multi-temporal segmentation labels and weights of the trainable layers are jointly optimized in iterations. We tested the method on a VHR dataset from Trento, Italy. Both quantitative and qualitative results demonstrated the effectiveness of the proposed approach.

Information from polarimetric synthetic aperture radar (PolSAR) imagery has been used for detecting built-up targets in classification problems, whereas it has been poorly exploited for change detection in multitemporal images. In this letter, we proposed an unsupervised approach for the detection of built-up changed areas from multitemporal full-polSAR images. The approach is based on the automatic thresholding of a novel change index based on the joint use of polarimetric span and average-alpha multitemporal information. The index is proposed for highlighting both constructed and demolished built-up elements. The experimental results on multitemporal UAVSAR images demonstrate that the proposed approach provides high detection accuracy and effectively separates among different types of changes, which is not the case with standard methods.

Due to the dramatically increased volume of remote sensing (RS) image archives, images are usually stored in a compressed format to reduce the storage size. Existing content-based RS image retrieval (CBIR) systems require as input fully decoded images, thus resulting in a computationally demanding task in the case of large-scale CBIR problems. To overcome this limitation, in this article, we present a novel CBIR system that achieves a coarse-to-fine progressive RS image description and retrieval in the partially decoded Joint Photographic Experts Group (JPEG) 2000 compressed domain. The proposed system initially: 1) decodes the code blocks associated only to the coarse wavelet resolution and 2) discards the most irrelevant images to the query image based on the similarities computed on the coarse resolution wavelet features of the query and archive images. Then, the code blocks associated with the subsequent resolution of the remaining images are decoded and the most irrelevant images are discarded by computing similarities considering the image features associated with both resolutions. This is achieved by using the pyramid match kernel similarity measure that assigns higher weights to the features associated with the finer wavelet resolution than to those related to the coarse wavelet resolution. These processes are iterated until the codestreams associated with the highest wavelet resolution are decoded. Then, the final retrieval is performed on a very small set of completely decoded images. Experimental results obtained on two benchmark archives of aerial images point out that the proposed system is much faster while providing a similar retrieval accuracy than the standard CBIR systems.

Satellite image time series (SITS), such as those by Sentinel-2 (S2) satellites, provides a large amount of information due to their combined temporal, spatial, and spectral resolutions. The high revisit frequency and spatial resolution of S2 result in: 1) increase in the probability of acquiring cloud-free images and 2) availability of detailed information for analyzing small objects. These characteristics are of interest in precision agriculture, where temporally dense SITS can benefit the understanding of crop behaviors. In the past, information about agricultural practices has been collected over large regions and focused on mixed/aggregated crops due to the poor tradeoff between the spatial and temporal resolutions. Products have been generated at low spatial resolution and daily basis or at high spatial resolution and weekly/monthly basis. They are meaningful for large agricultural fields, whereas they are limited when fields show a small average size. In this context, S2 characteristics allow for both high spatial and temporal resolution products. However, no existing automatic method effectively separates small fields from each other in an unsupervised way and deals with data irregularly sampled in time. Thus, this article presents a method suitable for the analysis of small crop fields in S2 dense SITS that accounts for S2 characteristics. The method fuses spatio-temporal information, analyzes data spatio-temporal evolution, and extracts relevant spatio-temporal information. The effectiveness of the proposed method was corroborated by experiments carried out on S2-SITS acquired over an area located in Barrax, Spain.

Very-high resolution (VHR) remote sensing images (RSIs) have significantly larger spatial size compared to typical natural images used in computer vision applications. Therefore, it is computationally unaffordable to train and test classifiers on these images at a full-size scale. Commonly used methodologies for semantic segmentation of RSIs perform training and prediction on cropped image patches. Thus, they have the limitation of failing to incorporate enough context information. In order to better exploit the correlations between ground objects, we propose a deep architecture with a two-stage multiscale training strategy that is tailored to the semantic segmentation of large-size VHR RSIs. In the first stage of the training strategy, a semantic embedding network is designed to learn high-level features from downscaled images covering a large area. In the second training stage, a local feature extraction network is designed to introduce low-level information from cropped image patches. The resulting training strategy is able to fuse complementary information learned from multiple levels to make predictions. Experimental results on two data sets show that it outperforms local-patch-based training models in terms of both accuracy and stability.

The possibility to apply passive radar sounding techniques to the Jovian icy moons, making use of Jupiter's strong decametric emissions (DAMs), has recently garnered large research interest. In this article, we propose a radar sounding simulation approach being able to simulate and compare passive and active acquisitions for a given scenario. The proposed simulator is based on the Stratton-Chu integral used with the linear phase approximation on triangular facets. The external field is modeled with plane waves, the direction, polarization, and amplitude of which can be freely chosen. The time-domain dependence can be either synthetically generated (e.g., Gaussian white noise) or taken from experimental measurements (e.g., waveforms recorded by a radio telescope observing Jupiter). For passive sounding, both autocorrelation and cross-correlation processing are considered. Validation tests were conducted on a series of ideal digital elevation models (DEMs), such as flat surfaces and subsurfaces, or Gaussian rough surfaces, and a good agreement with the existing literature was obtained. To illustrate the capabilities of the proposed simulator, we conducted additional more realistic experiments of radar sounding simulations, where we use both white noise and Jovian waveforms recorded by the LWA1 radio telescope.

The estimation of a snow-covered area (SCA) is often achieved by classification of imagery acquired by passive optical sensors aboard satellite platforms with high revisit frequencies [e.g., Moderate Resolution Imaging Spectroradiometer (MODIS)] required by various applications. The extraction of the SCA from optical imagery is inevitably hindered by the presence of clouds, where the surface labels (i.e., snow and no-snow) remain unknown. In the bulk of existing research, cloud pixels are either masked out without any further processing or assigned to snow/no-snow classes by performing spatial or temporal filters. The current approaches to deal with the cloud-obscuration problem are subjected to sizable uncertainties. They mostly neglect or only partially account for the temporal correlation, which undermines the full potential of long time series. We propose a novel method for estimating snow/no-snow labels beneath the clouds that leverages the multitemporal correlation between the presence/absence of snow and environmental factors including the topographical elevation, the date of acquisition (and thus the season), and the cloud-obscuration duration. The proposed method is built upon analyzing the long time series of maps derived from the single date classification of images in order to estimate the conditional probabilities of transition between the snow and no-snow classes. The probabilities are estimated as a function of the aforementioned environmental and multitemporal factors, which allow for the prediction of labels beneath the clouds in either archive or new acquisitions. Validation results on a four-year time series of daily MODIS images acquired over the Euregio region in Italian and Austrian Alps prove the effectiveness and robustness of the proposed method in assigning labels beneath the clouds.

Denoising is a common preprocessing step prior to the analysis and interpretation of hyperspectral images (HSIs). However, the vast majority of methods typically adopted for HSI denoising exploit architectures originally developed for grayscale or RGB images, exhibiting limitations when processing high-dimensional HSI data cubes. In particular, traditional methods do not take into account the high spectral correlation between adjacent bands in HSIs, which leads to unsatisfactory denoising performance as the rich spectral information present in HSIs is not fully exploited. To overcome this limitation, this article considers deep learning models-such as convolutional neural networks (CNNs)-to perform spectral-spatial HSI denoising. The proposed model, called HSI single denoising CNN (HSI-SDeCNN), efficiently takes into consideration both the spatial and spectral information contained in HSIs. Experimental results on both synthetic and real data demonstrate that the proposed HSI-SDeCNN outperforms other state-of-the-art HSI denoising methods. Source code: https://github.com/mhaut/HSI-SDeCNN.

Change detection (CD) is a crucial topic in many remote sensing applications. In the recent years, satellite polarimetric synthetic aperture radar (PolSAR) systems (e.g., the Sentinel-1 constellation) became a suitable tool for multitemporal monitoring due to the regular acquisitions with a short revisit time in different polarimetric channels. Methods for CD in PolSAR data mainly focus on binary CD (i.e., they provide information about the presence/absence of change only), whereas the polarimetric enhanced information provides multiple features that can be exploited for performing multiclass CD. In this article, we introduce a novel framework for the characterization of multitemporal changes in dual-polarimetric data. The framework is based on the definition of polarimetric change vectors (PCVs) and their representation in a polar coordinate system. PCVs allow characterizing and, thus, to separate multiclass changes in terms of target properties of the single-time scenes and the scattering theory. The proposed model is used to: 1) derive the statistical behaviors of change and no change classes in PolSAR multitemporal images; 2) design an automatic and unsupervised strategy to estimate the optimal number of changes; and 3) distinguish no change from change classes and the kinds of change from each other. An experimental analysis has been conducted on three multitemporal PolSAR data sets having different complexities in terms of number and kinds of change classes. The results confirm the effectiveness of the proposed approach and the better performance with respect to both specific techniques for CD in dual-pol SAR data and a general multiclass CD method, not designed for PolSAR data.

Recently, deep learning techniques have been introduced to address hyperspectral image (HSI) classification problems and have achieved the state-of-the-art performances. In this article, we propose a novel clustering algorithm for HSI based on learning embedding using the set-to-set and sample-to-sample distances (LSSDs). This technique consists of four main components: 1) oversegmentation; 2) generation of set-to-set and sample-to-sample distances; 3) learning embedding by training a siamese network; and 4) density-based spectral clustering. First, the HSI is oversegmented into superpixels by using the entropy rate superpixel (ERS) algorithm. Second, the set-to-set distances are obtained by representing the segmented sets of samples as affine hull (AH) models, whereas the sample-to-sample distances are computed by employing the local covariance matrix representation (LCMR) method. Third, sample pairs with the smallest and largest similarities are extracted according to the two distances. Then, these pairs are fed into the siamese multilayer perceptron (MLP) network and discriminative embeddings are learned by training the network with contrastive loss. Finally, density-based spectral clustering is applied to the deep embedding to obtain clustering results. Experimental results on three real HSIs demonstrate that the proposed method can achieve better performance than the considered baseline methods.

High/very-high-resolution (HR/VHR) multitemporal images are important in remote sensing to monitor the dynamics of the Earth's surface. Unsupervised object-based image analysis provides an effective solution to analyze such images. Image semantic segmentation assigns pixel labels from meaningful object groups and has been extensively studied in the context of single-image analysis, however not explored for multitemporal one. In this article, we propose to extend supervised semantic segmentation to the unsupervised joint semantic segmentation of multitemporal images. We propose a novel method that processes multitemporal images by separately feeding to a deep network comprising of trainable convolutional layers. The training process does not involve any external label, and segmentation labels are obtained from the argmax classification of the final layer. A novel loss function is used to detect object segments from individual images as well as establish a correspondence between distinct multitemporal segments. Multitemporal semantic labels and weights of the trainable layers are jointly optimized in iterations. We tested the method on three different HR/VHR data sets from Munich, Paris, and Trento, which shows the method to be effective. We further extended the proposed joint segmentation method for change detection (CD) and tested on a VHR multisensor data set from Trento.

Copernicus is the European programme for monitoring the Earth.It consists of a set of systems that collect data from satellites and in-situ sensors, process this data and provide users with reliable and up-to-date information on a range of environmental and security issues.The data and information processed and disseminated puts Copernicus at the forefront of the big data paradigm, giving rise to all relevant challenges, the so-called 5 Vs: volume, velocity, variety, veracity and value.In this short paper, we discuss the challenges of extracting information and knowledge from huge archives of Copernicus data.We propose to achieve this by scale-out distributed deep learning techniques that run on very big clusters offering virtual machines and GPUs.We also discuss the challenges of achieving scalability in the management of the extreme volumes of information and knowledge extracted from Copernicus data.The envisioned scientific and technical work will be carried out in the context of the H2020 project ExtremeEarth which starts in January 2019.

This paper addresses the opportunities and the challenges offered by multisource labeled data in the framework of deep learning techniques. After a review of the types of multisource labeled data, the focus is devoted to their use for the training of deep learning classification architectures. The need to generate training sets containing a very large number of labeled samples pushes toward the exploitation of all the possible available sources of labeled data. This crucial topic is addressed by categorizing the approaches to the collection of labeled data and presenting a framework for characterizing and modeling their information content and uncertainties to be used in the training of processing algorithms. The framework defines the main expected properties of large multisource training sets and relates them to both the characteristics of different data sources and the possible learning paradigms for the training of a deep architecture.

Crop-type classification has been attracting a lot of attention in recent years. In particular since the launch of the Sentinel-2 (S2) satellite which combines a large amount of spectral and spatial information, compared to previous satellite generations. In the literature, several methods exist that perform crop classification in time series, but most of them: i) work at pixel level; ii) perform single-data analysis; and/or iii) consider a single feature. This results in low performance of state-of-the-art methods. This paper presents an approach that works at object-level and exploits both spatial and temporal information coded in NDVI time series and phenological parameters and takes advantage of a semi-supervised paradigm by combining a new hierarchical correlation clustering with an artificial neural network. The effectiveness of the proposed approach was corroborated over an intensive cultivated area located in Barrax, Spain. Crop-type classification was compared to state-of-the-art methods.

Remote sensing images (RSIs) usually have much larger size compared to typical natural images used in computer vision applications. This makes the computational cost of training convolutional neural networks with full-size images unaffordable. Commonly used methodologies for semantic segmentation of RSIs perform training and prediction on cropped local image patches. Thus they fail to model the potential dependencies between ground objects at a higher level of abstraction. In order to better exploit global context information in RSIs, a deep architecture based on a two-stage training approach that is specially tailored to training large-size RSIs is proposed. In the first training stage, down-scaled images are used as input to learn high-level features from a large image area. In the second training stage, a local feature extraction network is designed to extract low-level information from cropped image patches. The complementary information learned from different levels is fused to make the prediction. As a result, the proposed two-stage training approach is able to exploit the context information of RSIs from a larger perspective without losing spatial details. Experimental results on a benchmark remote sensing dataset demonstrate the effectiveness of the proposed approach.

Radar sounders (RS) can be used to acquire data on ice sheets and provide direct evidence of the structures in the subsurface. Many acquisitions are available from airborne RS in the Antarctica and Greenland. However, airborne data are costly, have limited spatial coverage, and nonhomogeneous characteristics. To overcome these limitations, a potential satellite-mounted RS could provide uniform coverage and consistent data quality at the cost of lower resolution and higher path loss. In this paper, we assess the performance of a possible Earth-orbiting RS by simulating and analyzing its radargrams. The simulation approach reprocesses existing airborne RS to match the orbital RS characteristics. The simulated radargrams are analyzed to estimate the losses and understand the detection performance of icy targets using state-of-the-art data analysis techniques. The preliminary analysis of the simulated radargrams indicates that, under the simplified assumptions, an orbiting RS will be capable of imaging the investigated subsurface targets.

This paper introduces a novel method for estimation of snow/no-snow labels for cloud-obscured pixels in order to enable an accurate mapping of the snow-covered area (SCA) in time series. The proposed method leverages the embedded information in multitemporal correlation between the presence/absence of snow and environmental factors including the topographical elevation, date of acquisition, and the cloud obscuration duration. The proposed method is built upon three main steps: i) classification of single date images into three classes (snow, no-snow, and cloud), ii) estimation of conditional probabilities of class-transition in relation with the environmental factors, and iii) prediction of the snow/no-snow labels for the cloud-obscured pixels. We validated the proposed method on daily MODIS images acquired over 10 years in a mountain area located in Italy and Austria. The proposed method yielded SCA improved maps compared to a standard method of assigning labels beneath the clouds.

Individual tree detection in Light Detection and Ranging (LiDAR) data has been widely investigated in the literature. However, most of the methods work well on conifers but lead to poor accuracy in broad-leaved forest. The detection of deciduous trees is a complex task due to: (i) multiple local maxima present in the same canopy, and (ii) the tree-top (TP) can be in a different location from the canopy center. This paper presents an automatic technique which exploits high density LiDAR data to refine the detection of deciduous trees. First, the candidate tree-tops (CTPs) are detected using the standard level set method (LSM). Then, the Delaunay triangulation is used to generate a network topology which connects neighboring CTPs. For each pair of connected CTPs, geometrical features are extracted to automatically determine if the CTPs pair belongs to the same tree or to different canopies. The groups of CTPs identified as belonging to the same tree crown are merged into one TP. Preliminary numerical results show that the proposed method halves the commission errors of the initial TP detection by increasing the overall detection accuracy of 8.2%.

This paper proposes a novel approach to improve accuracy of weakly-supervised change detection in Synthetic Aperture Radar (SAR) imagery. The method is based on the idea to use an initial small size labeled dataset for synthetic training data generation (STDG) using an Ensemble of Self-Organizing Maps (ESOM). The resulting synthetic data set substitutes the initial small set of labeled authentic data and it is used to train a simple deep neural network (DNN) classifier. The proposed DNN architecture consists of 6 layers: an input layer, two Long Short-Term Memory (LSTM) layers, and three Fully Connected (FC) layers. The proposed method is evaluated on a TerraSAR-X image acquired in the Fukushima region, Japan, before and after tsunami. We have used only 200 labeled data (100 for the "change" class and 100 for the "no-change" class). As benchmark method, we have considered a Support Vector Machine (SVM) classifier. The experimental results confirmed the effectiveness of the proposed approach.

The accuracy in classification of remote sensing (RS) images using deep learning architectures is affected by the lack of large sets of training samples. Although a significant effort is currently devoted to generate databases of annotated satellite images, these datasets may not be large enough to accurately model at global level different types of land-cover surfaces. To solve such a problem, this paper presents an unsupervised approach which aims to exploit the RS image that has to be classified and publicly available thematic products to generate a training database of weak samples representative of the considered study area. First, we harmonize the thematic map and the RS image. Then, samples having the highest probability to be correctly associated to their labels are extracted from the map by exploiting the information provided by the RS image to be classified. Finally, the weak labeled samples are used to train a convolutional neural network (CNN). Experimental results obtained training a CNN on Sentinel 2 images with weak labels extracted from the 2018 corine land cover (CLC) map demonstrate the effectiveness of the proposed method.

The Italian Space Agency (ASI) has recently entrusted CO.RI.S.T.A. with the development of a radar system that can be mounted onboard small airplanes or helicopters and may operate, at different frequencies belonging to the P-Band, either as Synthetic Aperture Radar (SAR) or as Sounder. In this work, we present preliminary results of the helicopter-borne desert campaign carried out with this system in 2018 over the Erfoud area, Morocco, in the frame of a project that has involved different public Italian Research Institutes and Universities.

Change Detection (CD) using multitemporal satellite images is an important application of remote sensing. In this work, we propose a Convolutional-Neural-Network (CNN) based unsupervised multiple-change detection approach that simultaneously accounts for the high spatial correlation among pixels in Very High spatial Resolution (VHR) images and the differences in multisensor images. We accomplish this by learning in an unsupervised way a transcoding between multisensor multitemporal data by exploiting a cycle-consistent Generative Adversarial Network (CycleGAN) that consists of two generator CNN networks. After unsupervised training, one generator of the CycleGAN is used to mitigate multisensor differences, while the other is used as a feature extractor that enables the computation of multitemporal deep features. These features are then compared pixelwise to generate a change detection map. Changed pixels are then further analyzed based on multitemporal deep features for identifying different kind of changes (multiple-change detection). Results obtained on multisensor multitemporal dataset consisting of Quickbird and Pleiades images confirm the effectiveness of the proposed approach.

Radar for Icy Moon Exploration (RIME) is designed to characterize the subsurface geology of the Jovian icy moons. The RIME radargrams will show the combined response of a number of geophysical and geological characteristics of the ice-shells of these moons. Thus, radar sounder simulations are needed to understand the relationship between these target variables and the RIME response. In this paper, we use a computationally simple simulation approach that is based on reprocessing the radargrams available from the geological analogs of RIME targets. Moreover, we present a case study for a particular RIME target using this simulation technique for the generation of a database of RIME radar-grams, and a technique for analyzing this database. From the preliminary analysis of the simulated radargrams, we could derive important information regarding the underlying target variables. This confirms the usefulness of the presented approach to support the geological interpretation of the RIME radargrams.

Many research and industrial communities are betting on LoRa to provide reliable, long-range communication for the Internet of Things. This new radio technology, however, provides widely heterogeneous coverage; a LoRa link may span hundreds of meters or tens of kilometers, depending on the surrounding environment. This high variability is not captured by popular channel models for LoRa, and on-site measurements---a common alternative---are impractical due to the large geographical areas involved.

One of the most common approaches to unsupervised change detection (CD) in multispectral images is change vector analysis (CVA). CVA computes the multispectral difference image and exploits its statistical distribution in (hyper-) spherical coordinates by means of two steps: 1) magnitude and 2) direction thresholding. The two steps require assumptions on: 1) the model of class distributions and 2) the number of changes. However, both assumptions are seldom satisfied or difficult to formulate, especially when considering VHR images. Thus, we propose an approach to multiple CD in VHR optical images based on iterative clustering and adaptive thresholding in (hyper-) spherical coordinate. The proposed approach: 1) is distribution free; 2) is unsupervised; 3) automatically identifies the number of changes; and 4) is robust to noise. Results obtained on two multitemporal single-sensor and multisensor data sets, including images from WorldView-2 and QuickBird, corroborate the effectiveness of the proposed approach.

The repetitive acquisition of airborne light detection and ranging (LiDAR) data for forest surveys is rapidly increasing, thus making possible the forest dynamic analysis. Moreover, the availability of multitemporal data enables the possibility to improve the forest attribute estimates performed at single date, especially when one LiDAR acquisition has a lower pulse density with respect to the other. This letter presents a novel approach that exploits the bitemporal data information to: 1) improve the tree detection at both dates and 2) identify forest changes at single tree level. This is done by using a novel compound approach to the detection of trees in bitemporal data based on the Bayes rule for minimum error. Significant geometric features are extracted for each candidate tree-top and are used to estimate statistical terms employed in the compound approach. The multitemporal information is considered by estimating (in an iterative way) the probabilities of transition, which takes into account the temporal dependence between the LiDAR acquisitions. The proposed approach is evaluated on multitemporal LiDAR data acquired in a coniferous forest located in the Southern Italian Alps. Experimental results confirm the effectiveness of the compound detection that increases the overall accuracy (OA) up to 8.6% with respect to the single-date detection.

Hashing methods have recently attracted great attention for approximate nearest neighbor search in massive remote sensing (RS) image archives due to their computational and storage effectiveness. The existing hashing methods in RS represent each image with a single-hash code that is usually obtained by applying hash functions to global image representations. Such an approach may not optimally represent the complex information content of RS images. To overcome this problem, in this letter, we present a simple yet effective unsupervised method that represents each image with primitive-cluster sensitive multi-hash codes (each of which corresponds to a primitive present in the image). To this end, the proposed method consists of two main steps: 1) characterization of images by descriptors of primitive-sensitive clusters and 2) definition of multi-hash codes from the descriptors of the primitive-sensitive clusters. After obtaining multi-hash codes for each image, retrieval of images is achieved based on a multi-hash-code-matching scheme. Any hashing method that provides single-hash code can be embedded within the proposed method to provide primitive-sensitive multi-hash codes. Compared with state-of-the-art single-code hashing methods in RS, the proposed method achieves higher retrieval accuracy under the same retrieval time, and thus it is more efficient for operational applications.

Selection of threshold values to generate nonredundant filtered images in attribute profiles (APs) is an unresolved issue. This paper presents a novel filtering approach to the construction of APs that does not require the definition of any threshold value. The proposed approach creates a max-tree (or min-tree), traverse to the first encountered leaf node using depth first traversal, and defines a leaf attribute function (LAF) to demonstrate the changes in attribute values from leaf to root node. The LAF is analyzed based on a novel criterion to automatically detect the node along the path that has a first significant difference in the attribute value. All its descendant nodes are merged to it and the process is repeated for each unvisited leaf node to create the final filtered tree which is transformed back as a filtered image. The proposed approach can incorporate maximum spatial information by applying a few filtering operations without the need to define any threshold value. This is of great importance in spectral-spatial classification applications. Moreover, since the proposed approach requires one depth first traversal to generate a filtered image, it is very efficient in terms of computation time. To show the effectiveness of the proposed method, three real hyperspectral data sets are considered and the results are compared to the state-of-the-art method considering five different attributes. The results show that the proposed method has several important advantages with respect to the existing threshold-based filtering techniques. Furthermore, the proposed method is also effective when compared with different spectral-spatial classification techniques.

Spaceborne radar sounders are nadir-looking sensors operating in the high frequency (HF) or very high frequency (VHF) bands with subsurface sensing capabilities. Due to technological limitations, this type of sensors often deploys omnidirectional antennas. This results in undesired artifacts such as off-nadir clutter which could hinder data interpretation. Very recent technological advancements open up the possibility of synthesizing very large antenna apertures in HF/VHF band by using small satellites array deployed in suitable orbital formation flying. Accordingly, in this study, we propose a novel concept of distributed radar sounder system. The proposed concept is complemented with a mathematical model for performance prediction which takes into account the uncertainty on the position of the sensors. Moreover, we discuss possible orbital solutions for the problem of the deployment of the distributed radar sounder system. The results show that a distributed radar sounder operating in small satellites formation flying is particularly appealing as it can: 1) substantially reduce the impact of surface clutter; 2) increase the across-track resolution; 3) increase the signal-to-noise ratio (SNR) (or, alternatively, decrease the overall required transmitted power with respect to a traditional single configuration radar sounder design); and 4) provide large flexibility in the data processing of the signals acquired by the different sensors.

The availability of multitemporal images acquired by several very high geometrical resolution (VHR) optical sensors makes it possible to build VHR image time series (TS) of images acquired over the same geographical area with a temporal resolution better than the one achievable when considering a single VHR sensor. However, such TS include images showing different characteristics from the geometrical, radiometrical, and spectral viewpoint. Thus, there is a need of methods for building homogeneous VHR optical TS when using multispectral multisensor images. By focusing on the spectral domain, we propose a method to transform a VHR image into the spectral domain of another image in the same multisensor TS but acquired by a different sensor. To this end, a prediction-based approach relying on a nonparametric regression method is employed to mitigate sensor-dependent spectral differences. The impact of possible changes occurred on the ground is mitigated by training the prediction model on unchanged samples, only. Experimental results obtained on VHR optical multisensor images confirm the effectiveness of the proposed approach.

Accurate crown detection and delineation of dominant and subdominant trees are crucial for accurate inventorying of forests at the individual tree level. The state-of-the-art tree detection and crown delineation methods have good performance mostly with dominant trees, whereas exhibits a reduced accuracy when dealing with subdominant trees. In this paper, we propose a novel approach to accurately detect and delineate both the dominant and subdominant tree crowns in conifer-dominated multistoried forests using small footprint high-density airborne Light Detection and Ranging data. Here, 3-D candidate cloud segments delineated using a canopy height model segmentation technique are projected onto a novel 3-D space where both the dominant and subdominant tree crowns can be accurately detected and delineated. Tree crowns are detected using 2-D features derived from the projected data. The delineation of the crown is performed at the voxel level with the help of both the 2-D features and 3-D texture information derived from the cloud segment. The texture information is modeled by using 3-D Gray Level Co-occurrence Matrix. The performance evaluation was done on a set of six circular plots for which reference data are available. The high detection and delineation accuracies obtained over the state of the art prove the performance of the proposed method.

During the past decades, radar sounder (RS) instruments have been effectively used to detect subglacial lakes (SLs). SLs appear as flat, smooth, and bright reflectors in RS radargrams. The visual interpretation has been the main approach to SL detection in radargrams. However, this approach is subjective and inappropriate for processing large amounts of radargrams. While the analysis of RS data for understanding the subglacial hydrology has recently received increased attention, the literature on the development of automatic methods specifically designed for SL detection is still limited. In order to fill this gap, in this paper, we propose a novel automatic technique for SL detection. The technique is made up of two steps: 1) feature extraction and 2) automatic detection. In the first step, we define and extract three families of features for discriminating between the lake and nonlake radar reflections. The features model locally the basal topography, the shape of the basal reflected waveforms, and the statistical properties of the basal signal. In the second step, we provide the features as input to a support vector machine classifier to perform the automatic SL detection. The proposed technique has been applied to radargrams acquired over two large regions in East Antarctica and Siple Coast. The obtained results, which are validated both quantitatively and qualitatively, confirm the robustness of the features and their capabilities to effectively characterize SLs. Moreover, they prove the potentiality of the method to process large amounts of radargrams and update the current SL inventory.

The empirical (regression-based) methods for estimation of water quality parameters are mostly built upon the features derived from the original feature space of optical imagery (e.g., band ratios). This article aims at examining novel features to retrieve in-water constituents including chlorophyll-a (Chl-a), total suspended solids (TSS), and colored dissolved organic matter (CDOM). In this article, direction cosines and transformation of either color space or the coordinate system are applied to the original feature space in order to derive new features. A full-search approach is exploited to identify the optimal band combination for a given type of feature. The proposed analysis seeks for a band combination among all the possible ones that yield the strongest correlation through regressing a given feature against the concentration of the constituent of interest. The effectiveness of the proposed features is examined against standard ones using radiative transfer simulations, in situ measurements, and satellite imagery in a wide range of in-water optical conditions. The simulated and in situ data enabled in-depth analyses on the efficacy of recent satellite sensors with the primary focus of the aquatic science community [Operational Land Imager (OLI), MutiSpectral Instrument (MSI), and Ocean and Land Color Instrument (OLCI)] for retrieval of in-water constituents. TSS and Chl-a concentration of two alpine lakes (Lake Constance and Lake Lucerne) are also mapped using a real OLI image. The results suggest the effectiveness of the proposed features that can be leveraged to estimate the constituents in inland/coastal waters. OLI-based retrievals of in-water constituents proved difficulties in optically complex waters, whereas enhanced spectral resolution of MSI and OLCI permitted accurate estimates.

Hyperspectral (HS) images provide a dense sampling of target spectral signatures. Thus, they can be used in a multitemporal framework to detect and discriminate between different kinds of fine spectral change effectively. However, due to the complexity of the problem and the limited amount of multitemporal images and reference data, only a few works in the literature addressed change detection (CD) in HS images. In this paper, we present a novel method for unsupervised multiple CD in multitemporal HS images based on a discrete representation of the change information. Differently from the state-of-the-art methods, which address the high dimensionality of the data using band reduction or selection techniques, in this paper, we focus our attention on the representation and exploitation of the change information present in each band. After a band-by-band pixel-based subtraction of the multitemporal images, we define the hyperspectral change vectors (HCVs). The change information in the HCVs is then simplified. To this end, the radiometric information of each band is separately analyzed to generate a quantized discrete representation of the HCVs. This discrete representation is explored by considering the hierarchical nature of the changes in HS images. A tree representation is defined and used to discriminate between different kinds of change. The proposed method has been tested on a simulated data set and two real multitemporal data sets acquired by the Hyperion sensor over agricultural areas. Experimental results confirm that the discrete representation of the change information is effective when used for unsupervised CD in multitemporal HS data.

Change detection is one of the central problems in earth observation and was extensively investigated over recent decades. In this paper, we propose a novel recurrent convolutional neural network (ReCNN) architecture, which is trained to learn a joint spectral-spatial-temporal feature representation in a unified framework for change detection in multispectral images. To this end, we bring together a convolutional neural network and a recurrent neural network into one end-to-end network. The former is able to generate rich spectral-spatial feature representations, while the latter effectively analyzes temporal dependence in bitemporal images. In comparison with previous approaches to change detection, the proposed network architecture possesses three distinctive properties: 1) it is end-to-end trainable, in contrast to most existing methods whose components are separately trained or computed; 2) it naturally harnesses spatial information that has been proven to be beneficial to change detection task; and 3) it is capable of adaptively learning the temporal dependence between multitemporal images, unlike most of the algorithms that use fairly simple operation like image differencing or stacking. As far as we know, this is the first time that a recurrent convolutional network architecture has been proposed for multitemporal remote sensing image analysis. The proposed network is validated on real multispectral data sets. Both visual and quantitative analyses of the experimental results demonstrate competitive performance in the proposed mode.

Diameter at breast height (DBH) is one of the most important tree parameter for forest inventory. In this paper, we present a novel method for the adaptive and the accurate DBH estimation of trees characterized by small and large stems. The method automatically discriminates among different tree growth models by means of a data-driven technique based on a clustering procedure. First, the method detects young trees belonging to the lowest forest layer by simply considering the vertical structure of the forest. Then, different clusters of mature trees that are expected to share the same growth-model are identified by analyzing the environmental factors that can affect the stem expansion (e.g., topography and forest density). For each detected growth-model cluster, a tailored regression analysis is performed to obtain accurate DBH estimation results. Experiments have been carried out in an homogeneous coniferous forest located in the Alpine mountainous scenario characterized by a complex topography and a wide range of soil fertility. The method was tested on two data sets characterized by different light detection and ranging (LiDAR) point densities and different forest properties. The results obtained demonstrate the effectiveness of having multiple regression models adapted to the different growth models.

This paper aims to provide a compact superresolution formulation specific for multispectral (MS) multiresolution optical data, i.e., images characterized by different scales across different spectral bands. The proposed method, named multiresolution sharpening approach (MuSA), relies on the solution of an optimization problem tailored to the properties of those images. The superresolution problem is formulated as the minimization of an objective function containing a data-fitting term that models the blurs and downsamplings of the different bands and a patch-based regularizer that promotes image self-similarity guided by the geometric details provided by the high-resolution bands. By exploiting the approximately low-rank property of the MS data, the ill-posedness of the inverse problem in hand is strongly reduced, thus sharply improving its conditioning. The state-of-the-art color block-matching and 3D filtering (C-BM3D) image denoiser is used as a patch-based regularizer by leveraging the “plug-and-play” framework: the denoiser is plugged into the iterations of the alternating direction method of multipliers. The main novelties of the proposed method are: 1) the introduction of an observation model tailored to the specific properties of (MS) multiresolution images and 2) the exploitation of the high-spatial-resolution bands to guide the grouping step in the color block-matching and 3D filtering (C-BM3D) denoiser, which constitutes a form of regularization learned from the high-resolution channels. The results obtained on the real and synthetic Sentinel 2 data sets give an evidence of the effectiveness of the proposed approach.

This paper presents an unsupervised approach that extracts reliable labeled units from outdated maps to update them using time series (TS) of recent multispectral (MS) images. The method assumes that: 1) the source of the map is unknown and may be different from remote sensing data; 2) no ground truth is available; 3) the map is provided at polygon level, where the polygon label represents the dominant class; and 4) the map legend can be converted into a set of classes discriminable with the TS of images (i.e., no land-use classes that require manual analysis are considered). First, the outdated map is adapted to the spatial and spectral properties of the MS images. Then, the method identifies reliable labeled units in an unsupervised way by a two-step procedure: 1) a clustering analysis performed at polygon level to detect samples correctly associated to their labels and 2) a consistency analysis to discard polygons far from the distribution of the related land-cover class (i.e., having high probability of being mislabeled). Finally, the map is updated by classifying the recent TS of MS image with an ensemble of classifiers trained using only the reference data derived from the map. The experimental results obtained updating the 2012 Corine Land Cover (CLC) and the GlobLand30 in Trentino Alto Adige (Italy) achieved 93.2% and 93.3% overall accuracy (OA) on the validation data set. The method increased the OA up to 18% and 11.5% with respect to the reference methods on the 2012 CLC and the GlobLand30, respectively.

Efficient detection of targets immersed in a complex background with a low signal-to-clutter ratio (SCR) is very important in infrared search and tracking (IRST) applications. In this paper, we address the target detection problem in terms of local image segmentation and propose a novel small target detection algorithm derived from facet kernel and random walker (RW) algorithm which includes four main stages. First, since the RW algorithm is suitable for images with less noises, local order-statistic and mean filtering are applied to remove the pixel-sized noises with high brightness (PNHB) and smooth the infrared images. Second, the infrared image is filtered by the facet kernel to enhance the target pixels and candidate target pixels are extracted by an adaptive threshold operation. Third, inspired by the properties of infrared targets, a novel local contrast descriptor (NLCD) based on the RW algorithm is proposed to achieve clutter suppression and target enhancement. Then, the candidate target pixels are selected as central pixels to construct the local regions and the NLCD map of all local regions is computed. The obtained NLCD map is weighted by the filtered map of facet kernel to further enhance target. Finally, the target is detected by a thresholding operation on the weighted map. Experimental results on three data sets show that the proposed method outperforms conventional baseline methods in terms of target detection accuracy.

This paper presents a tensor alignment (TA) based domain adaptation (DA) method for hyperspectral image (HSI) classification. To be specific, HSIs in both domains are first segmented into superpixels, and tensors of both domains are constructed to include neighboring samples from a single superpixel. Then the subspace alignment (SA) between the two domains is achieved through alignment matrices, and the original tensors are projected as core tensors with lower dimensions into the invariant tensor subspace by applying projection matrices. To preserve the geometric information of original tensors, we employ a manifold regularization term for core tensors into the optimization process. The alignment matrices, projection matrices, and core tensors are solved in the framework of Tucker decomposition with an alternating optimization strategy. In addition, a postprocessing strategy is defined via pure samples extraction for each superpixel to further improve classification performance. Experimental results on four real HSIs demonstrate that the proposed method can achieve better performance compared with the state-of-the-art subspace learning methods when a limited amount of source labeled samples are available.

This paper introduces a novel heterogeneous domain adaptation (HDA) method for hyperspectral image (HSI) classification with a limited amount of labeled samples in both domains. The method is achieved in the way of cross-domain collaborative learning (CDCL), which is addressed via cluster canonical correlation analysis (C-CCA) and random walker (RW) algorithms. To be specific, the proposed CDCL method is an iterative process of three main components, i.e., RW-based pseudolabeling, cross-domain learning via C-CCA, and final classification based on extended RW (ERW) algorithm. First, given the initially labeled target samples as the training set (TS), the RW-based pseudolabeling is employed to update TS and extract target clusters (TCs) by fusing the segmentation results obtained by RW and ERW classifiers. Second, cross-domain learning via C-CCA is applied using labeled source samples and TCs. The unlabeled target samples are then classified with the estimated probability maps using the model trained in the projected correlation subspace. The newly estimated probability map and TS are used for updating TS again via RW-based pseudolabeling. Finally, when the iterative process converges, the result obtained by the ERW classifier using the final TS and estimated probability maps is regarded as the final classification map. Experimental results on four real HSIs demonstrate that the proposed method can achieve better performance compared with the state-of-the-art HDA and ERW methods.

Change detection (CD) in multitemporal images is an important application of remote sensing. Recent technological evolution provided very high spatial resolution (VHR) multitemporal optical satellite images showing high spatial correlation among pixels and requiring an effective modeling of spatial context to accurately capture change information. Here, we propose a novel unsupervised context-sensitive framework-deep change vector analysis (DCVA)-for CD in multitemporal VHR images that exploit convolutional neural network (CNN) features. To have an unsupervised system, DCVA starts from a suboptimal pretrained multilayered CNN for obtaining deep features that can model spatial relationship among neighboring pixels and thus complex objects. An automatic feature selection strategy is employed layerwise to select features emphasizing both high and low prior probability change information. Selected features from multiple layers are combined into a deep feature hypervector providing a multiscale scene representation. The use of the same pretrained CNN for semantic segmentation of single images enables us to obtain coherent multitemporal deep feature hypervectors that can be compared pixelwise to obtain deep change vectors that also model spatial context information. Deep change vectors are analyzed based on their magnitude to identify changed pixels. Then, deep change vectors corresponding to identified changed pixels are binarized to obtain a compressed binary deep change vectors that preserve information about the direction (kind) of change. Changed pixels are analyzed for multiple CD based on the binary features, thus implicitly using the spatial information. Experimental results on multitemporal data sets of Worldview-2, Pleiades, and Quickbird images confirm the effectiveness of the proposed method.

Simulation of radar sounder (RS) data is important for understanding the radar response of subsurface features to facilitate the interpretation of the real data. Conventional electromagnetic simulators require the definition of complex geoelectrical models of the investigated targets. They also involve time-complex solutions of Maxwell's equations for computing the received electric field, which leads to very high computation time. Furthermore, the simulated radargrams are often not realistic as it is very difficult to model all the variables involved in the data acquisition. In this paper, we propose a novel simulation approach that exploits the data available from existing RSs in geologically analogous terrains, to produce realistic simulations of the investigated RS target. This simulation strategy is based on minimizing the difference between the analog and the investigated acquisition scenarios. This is done by applying a series of corrections, which depend on the relation between the radargram characteristics and the physical variables describing the acquisition process. The aim is to produce radargrams that resemble the investigated scenario in terms of the echo magnitude, bandwidth, range resolution, and along-track resolution. Experimental results present three case studies for different possibilities of the analog and the investigated scenarios. The validation of the simulated radargrams with actual data demonstrates the effectiveness of the proposed approach. Finally, we also present a real application of this approach for the simulation of Radar for Icy Moon Exploration (RIME) radargrams for a combination of instrument and target parameters, using the SHAllow RADar (SHARAD) radargram acquired over the geological analog of a selected RIME target.

While image matching has been studied in remote sensing community for decades, matching multimodal data [e.g., optical, light detection and ranging (LiDAR), synthetic aperture radar (SAR), and map] remains a challenging problem because of significant nonlinear intensity differences between such data. To address this problem, we present a novel fast and robust template matching framework integrating local descriptors for multimodal images. First, a local descriptor [such as histogram of oriented gradient (HOG) and local self-similarity (LSS) or speeded-up robust feature (SURF)] is extracted at each pixel to form a pixelwise feature representation of an image. Then, we define a fast similarity measure based on the feature representation using the fast Fourier transform (FFT) in the frequency domain. A template matching strategy is employed to detect correspondences between images. In this procedure, we also propose a novel pixelwise feature representation using orientated gradients of images, which is named channel features of orientated gradients (CFOG). This novel feature is an extension of the pixelwise HOG descriptor with superior performance in image matching and computational efficiency. The major advantages of the proposed matching framework include: 1) structural similarity representation using the pixelwise feature description and 2) high computational efficiency due to the use of FFT. The proposed matching framework has been evaluated using many different types of multimodal images, and the results demonstrate its superior matching performance with respect to the state-of-the-art methods.

The deployment of a wireless sensor network (WSN) is crucial to its reliability and performance. Yet, node placement is typically determined in-field via effort-demanding trial-and-error procedures, because existing approaches over-simplify the radio environment; this especially holds for forests, the focus of this article, where trees greatly affect communication. We present LaPS (LiDAR-assisted Placement for wireless Sensor networks), an approach exploiting remote sensing to identify the best node placement automatically and prior to deployment . Airborne Light Detection and Ranging (LiDAR) data acquired for the target forest are automatically processed to estimate its properties (e.g., tree position and diameter) that, once incorporated into a specialized path loss model, enable per-link estimates of the radio signal attenuation induced by trees. Finally, a genetic algorithm explores placement options by evolving toward a (sub-)optimal solution while satisfying the user’s spatial and network requirements, whose formulation is very flexible and broadly applicable. Our experiments, focused on a real forest, confirm that LaPS yields topologies of significantly higher quality w.r.t. approaches using a regular placement or a standard path loss model. Further, the ability to quickly explore the impact that changes in user requirements have on topology is invaluable to improve the operation of WSNs and reduce the effort of their in-field deployment.

This paper presents a multitemporal technique for multi-class Change Detection (CD) between pairs of images of a satellite image time series. Changes between different pair of images within a time series must be consistent with each other since images acquired over the same scene are causally related with one another. The temporal consistency of the pixel status can be used to formulate a principle that constrains the CD results within the series to be mutually consistent. This principle coincides with the conservative property of the change variable and it allows the unsupervised validation of changes detected between arbitrary image pairs. Thus, all images in the series, rather than a single couple, are used in the pair-wise CD. The proposed technique was applied to a dataset of dual-polarized terrain-corrected SAR images acquired by Sentinel-1. Experimental results show the validity of the proposed multitemporal approach in improving the CD results.

In this work, we present a framework to integrate physically based hydroclimatological models and remote sensing products, by exploiting their advantages and overcoming their limitations, for an improved understanding and estimation of the alpine glacier accumulation and ablation processes. The capability of remote sensing to well represent the spatial variability of the snow cover over the glaciers is used to correct possible errors in the model simulations, thus obtaining a more reliable estimation of annual glacier mass balance. The proposed approach is tested on the glaciers in the Rofen Valley (Austria) by employing the AMUNDSEN model for accumulation and ablation processes simulation and Landsat-5/7/8 data for glacier zone mapping from 1998 to 2016.

Spaceborne radar sounders are nadir-looking sensors operating in HF/VHF bands. Their relatively low frequency allows the penetration of the transmitted signals into a given subsurface to infer its geoelectrical properties and composition by analyzing the radar echoes. These type of sensors are affected by unwanted artifacts such as off-nadir clutter mainly resulting from the constraint on the type of deployable antenna which is very often a dipole. Very recent technological advancements open up the possibility of synthesizing very large antenna apertures in HF/VHF band by using small satellites array deployed in suitable orbital formation flying. Accordingly, in this study we propose a novel concept of distributed radar sounder system. The proposed concept is complemented with a mathematical model to predict its performance. The results show that the distributed radar sounder operating in small satellites formation flying is particularly appealing as it can (i) drastically reduce the impact of surface clutter, (ii) increase the across-track resolution, and (iii) increase the signal to noise ratio (or, alternatively, decrease the overall required transmitted power with respect to a traditional radar sounder design).

Jupiter is one of the Solar System's most powerful source of radio waves in the MHz range. Very recent studies proposed that Jovian Decametric Radiation can be effectively exploited for probing the subsurface of Jupiter's icy moons Europa and Ganymede by using passive radio sounding. However, these studies results were based on the assumption that the Jovian noise is white and stationary. Therefore additional investigation into the temporal stability, predictability, and spectral properties of Jovian noise are required to fully assess the potential of passive radar sounding. In this paper, we investigate the actual spectral structure, temporal occurrence and general characteristics of the Jovian Decametric Radiation for understanding their impact on the radar sounding performance metrics. This is done by analyzing the recently available Jupiter's radio spectra acquired by JUNO Waves. The results show that the Jovian Decametric Radiation (i) is not perfectly white and some performance deterioration in the range response should be expected and (ii) has a varying temporal occurrence versus probing frequency.

Satellite Image Time Series (SITS), such as the ones acquired by the new Sentinel-2 (S2), combine a large amount of information compared to previous satellite generations since a better trade-off in terms of spatial/spectral/temporal resolutions is guaranteed. The specific characteristic of acquiring images under overlapped orbits, offered by S2, results in: i) availability of irregularly sampled acquisitions and ii) increase of the probability to acquire cloud free images over time. This characteristic becomes relevant in the agricultural analysis, where availability of dense SITS is required to map and analyze fast working crop behaviors. In the literature, several methods exist that extract phenological parameters for agricultural analysis, but none of them is able to deal with irregularly sampled data. Thus, this paper presents an approach for derivation of cropland phenological parameters from irregularly sampled S2-SITS. Experimental results obtained on S2-SITS acquired over Barrax, Spain, confirm the effectiveness of the proposed approach.

Lunar lava tubes are buried channels that contained thermally insulated lava during the volcanic period of the Moon. Nowadays, they are believed to be empty and thus, identified as potential habitats for humans. In recent years, numerous studies investigated the possible locations of these tubes by taking into account the distribution of gravity anomalies and the volcanic features of the surface. In this paper, we model lava tubes according to their electromagnetic behavior, and we propose a novel approach to locate lava tubes and estimate their physical properties. The method analyzes the subsurface reflections stored in radargrams to extract the desired features automatically. Then, these features and their relationships are processed by a fuzzy rule-based system to detect the presence or absence of lava tubes. The strategy was implemented and successfully tested on simulated radargrams with various surface properties and tunnel dimensions.

Reliable electromagnetic simulators are of prime importance for the design of radar sounder instruments and for supporting the subsequent interpretation of their data. In this paper we present a coherent simulator based on the facet method that can compute radar echoes from the subsurface of a target area with an arbitrary number of geological layers, thus going beyond the mere 1- or 2-layer descriptions usually modelled by coherent ray-tracing radar sounder simulators. The simulator has been validated using real radar data of lunar areas characterized by a multilayer nature collected by the Lunar Radar Sounder (LRS) instrument of JAXA's Kaguya probe. Results confirm the effectiveness of the proposed simulator.

The main idea of this study is the development of an innovative data fusion method through which state-of-the-art remotely sensed products and hydrological modelling simulations can be integrated to improve the retrieval and the reliability of snow cover and snow water equivalent mapping. The proposed method is based on a machine learning technique, Support Vector Machine (SVM), and on exploitation of two well-instrumented test-sites in EUREGIO region for the validation. Results show an improvement of performances with respect to single products from remote sensing and model. On EUREGIO scale the accuracy of snow cover mapping obtained from fusion reaches 0.95.

Approximately 70% of the total number of inventoried subglacial lakes (SLs) in Antarctica have been detected by visual interpretation or semiautomatic techniques applied to data acquired by airborne radar sounder (RS) instruments. Recently, interest has been shown in using automatic classifiers fed with topographic and structural features of the basal interface for the discrimination between lake and non-lake interfaces in RS data. To enhance the performances of the automatic classifiers, in this paper, we propose an additional set of three discriminant features of the basal interface. The features model the statistical properties of the basal reflected radar signal in terms of central moments and are particularly suitable to the accurate description of subglacial lakes since they i) locally characterize the basal interface, ii) do not rely on subsurface attenuation models, and ii) are independent on depth. The effectiveness of the proposed statistical features has been proven experimentally using a large RS dataset acquired in East Antarctica.

This paper focuses on solving the multi-class change detection problem in bitemporal multispectral remote sensing images. In that case, information that represented in a small number (e.g., two) of the original bands may be insufficient for the accurate identification of a few of multi-class changes. In particular, this problem becomes more difficult in unsupervised change detection cases when ground reference data is not available. In this paper, a solution is proposed by using the potential information represented in expanded features that constructed from the original spectral bands. Experimental results obtained on a real bitemporal remote sensing data set confirm the effectiveness of the proposed approach.

In this paper, a novel approach for the retrieval of snow wetness is presented for Sentinel-1 (S-1) data. The approach uses the information on snow proprieties provided by the hydroclimatological model AMUNDSEN and confirmed by comparisons performed at different sematic level to train a regressor that is able to exploit the dual-polarimetric information provided by S-1. The preliminary results obtained for the Rofental in Austria are discussed.

This paper presents a Change Detection method for bitemporal Light Detection And Ranging (LiDAR) data based on Change Vector Analysis in the polar domain. The method first extracts a suite of LiDAR metrics from the two LiDAR point clouds using a 2-D grid based approach. Second, it transforms the change in these metrics into a polar representation to examine variations in terms of magnitude and direction. The analysis of the magnitude discriminates between small magnitude changes or unchanged areas and areas affected by large disturbances related to forest removal. The analysis of the direction of change allows us to identify dominant directions to discriminate between the various types of forest change. The method has been tested on a multitemporal dataset acquired in a high productivity evergreen conifer forest in British Columbia, Canada. Experimental results indicated that the method effectively discriminates between the different types of forest change trough the analysis of the change direction.

The increasingly availability of Light Detection and Ranging (LiDAR) data acquired at different times can be used to analyze the forest dynamics at individual tree level. This often requires to deal with LiDAR point clouds having significantly different point densities. To address this issue, this paper presents a method for the fusion of multitemporal Li-DAR data which aims at using the information provided by high density LiDAR data (higher than 10 pts/m2) to improve the single tree parameter estimation of low density data (up to 5 pts/m2) acquired over the same forest at different times. The method first accurately characterizes the crown shapes on the high density data. Then, it uses the obtained estimates to drive the tree parameter estimation on the low density LiDAR data. The method has been tested on a multitemporal dataset acquired in coniferous forests located in the Italian Alps. Experimental results confirmed the effectiveness of the method.

Combining existing thematic vector products and recently acquired satellite images to generate regular updated maps is extremely interesting at operational level. However, employing these maps is not straightforward. They are typically provided at polygon level, where the polygon labels do not necessarily correspond to spectrally homogeneous areas. Moreover, usually there is a semantic gap between the map legend and the set of natural classes discriminable in multispectral images. To overcome these issues, this paper presents a method that first performs a domain understanding to detect the discrepancies between the vector map domain and the multispectral (MS) image domain. Then, it accomplishes a domain modeling which uses a MS image contemporary to the map to extract a set of reliable and informative samples from the map. Finally, the method carries out Domain Adaptation (DA) using a recent MS image to update the map. Experimental results obtained updating a crop thematic map in Czech Republic confirm the effectiveness of the method.

Change Detection (CD) using multi-temporal satellite images is a fundamental application of remote sensing. To effectively capture change information from Very High spatial Resolution (VHR) optical images, spatial context needs to be modelled as VHR images are characterized by high spatial correlation among pixels. We propose a context-sensitive framework for CD in multitemporal VHR images using pre-trained Convolutional-Neural-Network (CNN)-based feature extraction. Such a framework, while unsupervised, can effectively model the spatial relationship among neighbouring pixels in VHR images. A CNN, pre-trained for semantic segmentation, enables us to obtain multi-temporal deep features that are compared pixelwise to identify changed pixels. Changed pixels are further clustered for multiple change detection. Results obtained on multi-temporal datasets of Worldview-2 and Pleiades images demonstrate effectiveness of our approach.

We propose a system for detecting clear-cuts in Sentinel-2 (S-2) images of the Indonesian forest by means of an adaptive and unsupervised multivariate Change Vector Analysis (CVA) method. By leveraging on the unique spatial and spectral characteristics of the S- 2 mission, the proposed method characterizes a relevant portion of the target change as lying in a Gaussian neighborhood of the spectral stacked bi-temporal domain of the change. The processing system analyzes all the available bi-temporal pairs in the time series, enabling us to: (1) partially recovering lost information due to cloud coverage, and (2) providing a representation of the change evolving in time. The system is fully automated and potentially operational ready, so it can be used to provide accurate information about clear-cuts at the country scale in Indonesia.

Spaceborne low-frequency and wide bandwidth radar sounders are a promising technology to regularly investigate at global-scale Earth's icy and arid regions. However, Earth ionosphere distorts the radar signal impacting performance parameters, such as subsurface resolution, of the radar system. One of the most relevant distortions that a sounder signal in the lower part of the very high-frequency (VHF) band (e.g., 40-50 MHz) encounters is the distortion of the phase component that could become mission critical if not properly compensated. Low-frequency and high fractional bandwidth radar systems are particularly affected by this issue. Previous works on radar sounder ionosphere phase distortion compensation addressed the Martian ionosphere and used techniques based on the Taylor series expansion. In this letter, we focus on the Earth ionosphere and we exploit a recently proposed ionosphere compensation technique based on the Legendre orthogonal polynomials expansion, which proved to be more accurate than the compensation based on Taylor expansion. Simulations show that the method allows a nominal compensation of the phase distortions under realistic ionosphere scenarios expected during the acquisitions. Furthermore, it proved to be accurate and robust for total electron content conditions expected during nighttime for all the geomagnetic latitudes. The results confirm that the method can accurately compensate the distorting effects on the phase component of a spaceborne VHF radar sounder.

This paper presents an unsupervised approach to band selection in hyperspectral images that considers both spectral and spatial information in data dimensionality reduction. The approach exploits the concepts of superpixel and chunklets for identifying the spectral channels most suitable to be used in classification for discriminating land-cover classes. The segmented superpixels can be regarded as many small spectral homogeneous and spatial neighboring pixel chunklets. Based on the observation that the superpixel chunklets achieve high homogeneity and consistency within land-cover classes, a series of band criteria (BC) is identified by learning the optimal band transformation that results in low within-class variability and high total variability. Then, the learned BC, which are called band measures, are given in input to an efficient clustering algorithm, i.e., the affinity propagation, for selecting highly separable bands with low redundancy. The effectiveness of proposed approach was assessed on three hyperspectral data sets. The results point out the advantages of the proposed methods over five state-of-the-art unsupervised methods.

Conventional supervised content-based remote sensing (RS) image retrieval systems require a large number of already annotated images to train a classifier for obtaining high retrieval accuracy. Most systems assume that each training image is annotated by a single label associated to the most significant semantic content of the image. However, this assumption does not fit well with the complexity of RS images, where an image might have multiple land-cover classes (i.e., multilabels). Moreover, annotating images with multilabels is costly and time consuming. To address these issues, in this paper, we introduce a semisupervised graph-theoretic method in the framework of multilabel RS image retrieval problems. The proposed method is based on four main steps. The first step segments each image in the archive and extracts the features of each region. The second step constructs an image neighborhood graph and uses a correlated label propagation algorithm to automatically assign a set of labels to each image in the archive by exploiting only a small number of training images annotated with multilabels. The third step associates class labels with image regions by a novel region labeling strategy, whereas the final step retrieves the images similar to a given query image by a subgraph matching strategy. Experiments carried out on an archive of aerial images show the effectiveness of the proposed method when compared with the state-of-the-art RS content-based image retrieval methods.

Change detection from multitemporal synthetic aperture radar (SAR) images enables mapping applications for earth environmental observation, human activity monitoring, and urban studies. We expand the use of SAR data beyond single-look processing to include the spatial response of targets. This information is derived from a multisquint framework similar to beamforming. To preserve changes detected at nominal resolution, a three-stage change detector exploiting single-look and multisquint processing mode is proposed to mitigate false alarms caused by image artifacts typically found in high-resolution SAR imagery and urban scenarios. After applying the proposed method to multitemporal images, the false alarm rate was reduced by a factor 3, while preserving 95% of the detection rate offered by traditional schemes.

Reliable electromagnetic simulators are of prime importance for the design of radar sounder instruments and for supporting the subsequent analysis of their data. In this paper, we present a coherent, facet method-based simulator that can compute radar echoes from the subsurface of a target area with an arbitrary number of geological layers, thus going beyond the surface-only or the two-layer descriptions so far implemented in coherent ray-tracing radar sounder simulators. Propagation of fields throughout the subsurface is computed according to Snell's law following a ray-tracing approach. For each ray interacting with the surface, be it a direct reflection or a refracted ray coming from the subsurface, the phase contribution of each facet is calculated through the linear phase approximation, while the total field received at the antenna is computed using Huygen's principle. Validation simulations have been performed against the radar data of lunar and martian areas characterized by a multilayer nature and collected by the Lunar Radar Sounder instrument of JAXA's Kaguya lunar probe and the Shallow Radar instrument onboard NASA's Mars Reconnaissance Orbiter, respectively. Results confirm the effectiveness of the proposed simulator.

The supervised deep networks have shown great potential in improving the classification performance. However, training these supervised deep networks is very challenging for hyperspectral image given the fact that usually only a small amount of labeled samples are available. In order to overcome this problem and enhance the discriminative ability of the network, in this paper, we propose a deep network architecture for a super-resolution (SR)-aided hyperspectral image classification with classwise loss (SRCL). First, a three-layer SR convolutional neural network (SRCNN) is employed to reconstruct a high-resolution image from a low-resolution image. Second, an unsupervised triplet-pipeline CNN (TCNN) with an improved classwise loss is built to encourage intraclass similarity and interclass dissimilarity. Finally, SRCNN, TCNN, and a classification module are integrated to define the SRCL, which can be fine-tuned in an end-to-end manner with a small amount of training data. Experimental results on real hyperspectral images demonstrate that the proposed SRCL approach outperforms other state-of-the-art classification methods, especially for the task in which only a small amount of training data are available.

Most traditional approaches classify hyperspectral image (HSI) pixels relying only on the spectral values of the input channels. However, the spatial context around a pixel is also very important and can enhance the classification performance. In order to effectively exploit and fuse both the spatial context and spectral structure, we propose a novel two-stream deep architecture for HSI classification. The proposed method consists of a two-stream architecture and a novel fusion scheme. In the two-stream architecture, one stream employs the stacked denoising autoencoder to encode the spectral values of each input pixel, and the other stream takes as input the corresponding image patch and deep convolutional neural networks are employed to process the image patch. In the fusion scheme, the prediction probabilities from two streams are fused by adaptive class-specific weights, which can be obtained by a fully connected layer. Finally, a weight regularizer is added to the loss function to alleviate the overfitting of the class-specific fusion weights. Experimental results on real HSIs demonstrate that the proposed two-stream deep architecture can achieve competitive performance compared with the state-of-the-art methods.

Light Detection and Ranging (LiDAR) data have been widely used to characterize the 3-D structure of the forest. However, their use in a multitemporal framework has been quite limited due to the relevant challenges introduced by the comparison of pairs of point clouds. Because of the irregular sampling of the laser scanner and the complex structure of forest areas, it is not possible to perform a point-to-point comparison between the two data. To overcome these challenges, a novel hierarchical approach to the detection of 3-D changes in forest areas is proposed. The method first detects the large changes (e.g., cut trees) by comparing the Canopy Height Models derived from the two LiDAR data. Then, according to an object-based change detection approach, it identifies the single-tree changes by monitoring both the treetop and the crown volume growth. The proposed approach can compare LiDAR data with significantly different pulse densities, thus allowing the use of many data available in real applications. Experimental results pointed out that the method can accurately detect large changes, exhibiting a low rate of false and missed alarms. Moreover, it can detect changes in terms of single-tree growth, which are consistent with the expected growth rates of the considered areas.

This paper presents multiple kernel learning (MKL) in the context of remote sensing (RS) image classification problems by illustrating main characteristics of different MKL algorithms and analyzing their properties in RS domain. A categorization of different MKL algorithms is initially introduced, and some promising MKL algorithms for each category are presented. In particular, MKL algorithms presented only in machine learning are introduced in RS. Then, the investigated MKL algorithms are theoretically compared in terms of their: 1) computational complexities; 2) accuracy with different qualities of kernels; and 3) accuracy with different numbers of kernels. After the theoretical comparison, experimental analyses are carried out to compare different MKL algorithms in terms of: 1) model selection and 2) feature fusion problems. On the basis of the theoretical and experimental analyses of MKL algorithms, some guidelines for a proper selection of the MKL algorithms are derived.

This paper presents a sensor-driven hierarchical domain adaptation method that aims at transferring the knowledge from a source domain (RS image where reference data are available) to a different but related target domain (RS image where no labeled reference data are available) for solving a classification problem. Due to the different acquisition conditions, a difference in the source and target distributions of the features representing the same class is generally expected. To solve this problem, the proposed method takes advantage from the availability of multisensor data to hierarchically detect features subspaces where for some classes data manifolds are partially (or completely) aligned. These feature subspaces are associated with invariant physical properties of classes measured by the sensors in the scene, i.e., measures having almost the same behavior in both domains. The detection of these invariant feature subspaces allows us to infer labels of the target samples that result more aligned to the source data for the considered subset of classes. Then, the labeled target samples are analyzed in the full feature space to classify the remaining target samples of the same classes. Finally, for those classes for which none of the sensors can measure invariant features, we perform the adaptation via a standard active learning technique. Experimental results obtained on two real multisensor data sets confirm the effectiveness of the proposed method.

The change detection (CD) problem is very important in the remote sensing domain. The advent of a new generation of multispectral (MS) sensors has given rise to new challenges in the development of automatic CD techniques. In particular, typical approaches to CD are not able to well model and properly exploit the increased radiometric resolution characterizing new data as this results in a higher sensitivity to the number of natural classes that can be statistically modeled in the images. In this paper, we introduce a theoretical framework for the description of the statistical distribution of the difference image as a compound model where each class is determined by temporally correlated class transitions in the bitemporal images. The potential of the proposed framework is demonstrated on the very common problem of binary CD based on setting a threshold on the magnitude of the difference image. Here, under some simplifying assumptions, a multiclass distribution of the magnitude feature is derived and an unsupervised method based on the expectation-maximization algorithm and Bayes decision is proposed. Its effectiveness is demonstrated on a large variety of data sets from different MS sensors. In particular, experimental tests confirm that: 1) the fitting of the magnitude distribution significantly improves if compared with already existing models and 2) the overall CD error is close to the optimal value.

Change detection (CD) between a pair of images is a popular problem in remote sensing. Despite a large amount of data is acquired every day by remote sensing satellites, standard CD methods usually consider only the two target images between which we desire to detect changes. The aim of this work is to present a novel framework in which the bi-temporal CD is redefined by evaluating the consistency of the changes occurred in the target image pair with all the other changes of images within the considered time series. Our approach evaluates pixel-wise the changes in temporal closed-loops that include the two target images where the resulting binary change/no-change sequences can be processed by strategies inspired to the error-control-coding theory. Unreliable CD results for the target images can be identified and corrected. The experimental results on both a synthetic and a real dataset demonstrate the effectiveness of the proposed framework.

This paper presents a novel content based remote sensing (RS) image retrieval system that consists of: i) a spatial and spectral image description scheme; and ii) a sparsity based supervised retrieval method. Spatial image description is based on the scale invariant feature transform (SIFT), while a novel descriptor defined based on the bag of spectral values is proposed to express spectral features. With the conjunction of these two feature vectors RS image retrieval is instrumented via a sparse reconstruction-based approach. These sparse reconstructions are used to estimate the likelihood of a scene to contain a land-cover class label. Applying this method separately for each land-cover class, one achieves retrieval in the framework of multi-label remote sensing image retrieval. Experimental results obtained on an archive of hyperspectral images show the effectiveness of the proposed system.

Urban change detection is an important part of monitoring operations and disaster relief efforts. However, often sufficient ground truth data is not available to use traditional supervised machine learning techniques. In this paper, a novel Deep Learning based weakly-supervised framework for urban change detection using multi-temporal polarimetric SAR data is proposed. A modified unsupervised stacked auto-encoder stage is used to learn an efficient representation of the multi-temporal polarimetric information. Then a label aggregation is performed in the feature space before classification by a multi-layer perceptron. The proposed methodology is validated on a L-band UAVSAR dataset acquired over Los Angeles, CA and performs accurately and effectively with a low false alarm rate.

Airborne Light Detection and Ranging (LIDAR) remote sensing based forest inventory at the individual tree level is a valuable and effective alternative to manual inventory, due to factors such as higher accuracy, easy repeatability of sampling, and economic benefits. However, individual tree detection in multi-storied forests is challenging due to high tree proximity and forest structure complexity issues. In this work, we aim at detecting subdominant trees in a multi-stored forest from high density small foot-print multi-return airborne LiDAR data. The marker controlled watershed segmentation is used for the three dimensional (3D) delineation of the dominant tree crowns. The data associated with every segment are separately projected onto a novel 3D space, where crown surface information is effectively represented and subdominant trees are highlighted. A set of ten features is employed to separate subdominant from dominant trees. Preliminary results prove the effectiveness of the proposed method.

The automatic analysis of radargrams acquired by radar sounder instruments is an important task, as it has been often highlighted by the scientific community. In particular, an automatic classification technique enables a fast and objective extraction of ice subsurface target properties on wide areas. The main drawback of the available classification techniques is the use of fixed size windows for feature extraction, which causes a poor classification performance at the transition boundary between adjacent subsurface targets. To address this issue, in this paper, we propose a novel technique for the automatic classification of ice subsurface targets in radargrams. The method provides two key contributions, i.e., i) the use of a window size estimation method to adaptively characterize the subsurface targets, and ii) the use of the estimated windows for extracting novel discriminant features for automatic classification. Quantitative and qualitative results obtained by applying the method to a data set of radargrams acquired by an airborne-mounted radar sounder instrument in central Antarctica show the effectiveness of the proposed technique, which outperforms previously proposed methods.

The empirical line (EL) calibration is commonly used for atmospheric correction of remotely sensed spectral images and recovery of surface reflectance. Current methods for EL calibration are applied to single image using two (or more) reference targets. Considering cases with large number of (partially overlapped) images, only few scenes will include reference targets. Moreover, applying the estimated calibration coefficients of one image to other images can cause wrong results. Accordingly, the use of EL calibration is impractical for these cases. In this paper, we present a novel method for a simultaneous calibration of multiple images, which is called multiple image constrained empirical line (MIcEL). We present a generalized EL model that provide constrained results and is adaptable for large number of images. Given a set of images, we use available reference targets and tie points between overlapping images to calibrate all the images in the set simultaneously. Tie points are automatically extracted using scale-invariant feature transform (SIFT) method. Accuracy assessment of the MIcEL was carried out using real hyperspectral images and field measurements. The performance of MIcEL was compared to the quick atmospheric correction (QUAC) method. the results show that (comparable with respect to QUAC) the absolute accuracy of the MIcEL, with respect to filed measurements, is ~ ± 11%.

A novel spectral-spatial joint multiscale approach is developed to address the multi-class change detection problem in bitemporal multispectral remote sensing images. The proposed approach is based on a multiscale morphological compressed change vector analysis (M2C2VA), which extend the state-of-the-art spectrum-based compressed change vector analysis (C2VA) while preserving more geometrical details of change targets. In particular, spectral change features are reconstructed according to the morphological analysis which exploiting the interaction of a pixel with its adjacent regions. Two multiscale ensemble strategies are proposed to integrate the change information represented at multiple scales in order to enhance the CD performance. The proposed approach is designed in an unsupervised fashion thus can be implemented without using ground reference data. A pair of real bitemporal remote sensing images is used to test the proposed approach and the obtained experimental results confirm its effectiveness.

This paper presents a novel semisupervised framework for detecting multi-class changes in bitemporal hyperspectral images. By taking advantages of the state-of-the-art unsupervised change representation technique and the advanced supervised classifiers, the proposed framework allows the generation of pseudo training samples associated with the no-change and each change class that learned from the multitemporal data and import them into the supervised classifiers. Thus multiple changes can be discriminated from the original or the transformed feature space. The proposed approach was validated on a pair of real bitemporal Hyperion hyperspectral images, and the obtained experimental results confirm its effectiveness in addressing the challenging multi-class change detection task in hyperspectral images.

Multitemporal Hyperspectral (HS) images can be used in Change Detection (CD) to identify and discriminate among different kinds of change due to the fine sampling of the spectrum by HS sensors. In this work we propose a novel method for unsupervised multiple CD in multitemporal HS data based on binary Spectral Change Vectors (SCVs) and an agglomerative hierarchical clustering. First, we perform binary CD to separate changed from unchanged pixels. Second, we convert the real valued SCVs into binary ones. Thus we move from a real valued high dimensional space to a discrete one. The binary signatures are used to construct a dendrogram following an hierarchical agglomerative clustering approach. Finally, we exploit the hierarchical structure to discriminate among the kinds of change in a fully unsupervised manner. The experimental results obtained on the real dataset confirmed the effectiveness of the proposed method.

In this paper, we focus the attention on the superresolution of multispectral (MS) multiresolution images (e.g., Sentinel 2, Aster, MODIS). By taking advantage of the high spatial resolution bands, we minimize an objective function containing a quadratic data fitting term, an edge preserving regularizer, and a patch-based plug-and play prior promoting self-similar images. To cope with the ill-posedness of the problem we i) exploit the fact that the images are approximately low-rank, and ii) propose a hierarchical method which sharpens in the first place the medium resolution bands and then the coarse resolution ones. The optimization is solved with the alternating direction method of multipliers (ADMM), yielding a fast, flexible, and effective solver, named Superresolution MUltiband multireSolution Hierarchical approach (SMUSH). Quantitative and qualitative results obtained on simulated and real Sentinel 2 (S2) images show the SMUSH effectiveness.

This paper presents an approach to the update of land-cover maps by classifying Remote Sensing (RS) images in an unsupervised way. The proposed method assumes that: i) an old thematic map is available; ii) no ground truth data are available; iii) the source used to generate the available thematic map is unknown. To classify the most recent RS image available on the considered area, the method automatically extracts from the considered land-cover map a “pseudo” training set. First, a preprocessing phase adapts the map to the properties of the RS data. Then, we perform an automatic “pseudo” training set identification to select the most reliable samples from the existing thematic map. Finally, a consistency check is defined to determine whether the inconsistencies between the updated and the original maps are due to real changes on the ground or classification errors. Experimental results obtained by updating the 2012 Corine Land Cover Map (CLC) in Trentino, Italy, using Sentinel 2 (S2) images confirm the effectiveness of the proposed method.

Change detection in large urban areas is an application with increasing relevance. In this domain, Polarimetric SAR (PolSAR) sensors are receiving more attention recently. The enhanced polarimetric information provides useful features which can describe multi-temporal changes. In this work, we aim at introducing an approach for unsupervised change detection with focus on built-up areas that relies on the polarimetric information. This approach is based on the analysis of the multi-temporal α feature obtained from the Cloude-Pottier eigenvalue/eigenvector decomposition. Large differences in the α values can be associated to changes in the dominant scattering mechanism. These are likely to be associated to buildings when built-up areas are considered. Changes are detected according to an automatic and unsupervised approach. Validation is conducted on a pair of UAVSAR images acquired over Los Angeles, USA. Preliminary results highlight the effectiveness of proposed approach.

This paper proposes a novel unsupervised method based on primitive cluster sensitive hashing for fast and accurate image retrieval in large remote sensing (RS) archives. The proposed method consists of a three-steps algorithm. In the first step, each image in the archive is characterized by primitive clusters' descriptors. These descriptors are obtained through an unsupervised approach, which automatically extracts the image regions' descriptors and then associates them with primitive clusters. In the second step the primitive clusters' descriptors are transformed into multi-hash codes to represent each image. Then, in the last step, a multi-hash-code-matching scheme is applied to retrieve the images in the archive that are very similar to a query image. Experiments carried out on an archive of aerial images show that the proposed method provides distinctive multi-hash codes associated to the primitive clusters. Thus, it is more accurate than standard hashing methods, particularly under complex RS image retrieval tasks.

This paper presents a fast and robust framework integrating local features for the matching of multimodal geospatial data (e.g., optical, LiDAR, SAR and map). In the proposed framework, local feature descriptors, such as Histogram of Oriented Gradient (HOG) and Local Self Similarity (LSS), are first extracted for every pixel to form a pixel-wise structural feature representation of an image. Then we define a similarity metric based on the feature representation in frequency domain using the 3 Dimensional Fast Fourier Transform (3DFFT) technique, followed by a template matching scheme to detect control points between multimodal data. The proposed framework is based on the hypothesis that structural similarity between images is preserved across different modalities. The major advantages of this framework include (1) structural similarity representation using pixel-wise feature description and (2) high computational efficiency due to the use of 3DFFT. Experimental results on different types of multimodal geospatial data show more accurate matching performance of the proposed framework than the state-of-the-art methods.

The increased radiometric resolution of last generation multispectral sensors results in large statistical variability of classes represented in the image. However, classes present high spatial homogeneity. To preserve classes identity while simplifying their representation, in this paper we propose a class-wise spatial-contextual method based on a variational model with free discontinuities that reduces the statistical variability of classes by emphasizing their spatial contours. To prove its effectiveness, the proposed method is applied in the context of change detection in multispectral images. Here, it is able to augment the discrimination between the unchange and the change classes and to improve the detection performance.

Radar sounders are unique instruments for subsurface investigation in both terrestrial and space applications. They are widely employed for monitoring changes to the polar ice sheets and for the study of planetary bodies (e.g., Mars). The analysis of the very large amount of data produced by such systems requires the development of automatic techniques for an objective, accurate, and fast extraction of relevant information from radargrams. In this paper, we propose a novel technique for the automatic detection of layer boundaries based on a local scale hidden Markov model (HMM), which models the radar response in the presence of a layer boundary, and the Viterbi algorithm (VA, which performs the inference step). The proposed technique is based on a divide and conquer strategy that executes the VA using the observation data and the HMM to infer the most likely layer boundary location within a small radargram portion. Finally, a detection strategy is defined to chain together the inferred local layer locations. Furthermore, a novel radargram enhancement and denoising technique tailored to support the detection step is presented. The effectiveness of the proposed technique has been confirmed by testing it on different radargrams acquired by shallow radar over the north pole of Mars. The results obtained point out the superiority of the proposed method in retrieving the position of each layer boundary (and thus of the related intensity and geometric properties) with respect to the state-of-the-art techniques.

Surface cross-track clutter can corrupt both earth and planetary radar sounder (RS) observations preventing definitive interpretation of subsurface features, which are often of primary interest to geologists and planetary scientists. This clutter is usually identified either by manual or automatic techniques that require ancillary information about the topography of the surface, or by using multichannel RS systems with arrays of antennas. However, topographic information is not always available and multichannel systems are generally too massive and costly to mount on satellites for the planetary exploration. In this paper, we propose a novel approach to clutter discrimination that is independent of ancillary information and limits the hardware complexity of the RS system. This approach uses a two-channel RS and exploits cross-channel interferometric phase differences to discriminate the clutter. Our approach includes three main steps: 1) manual feature extraction and theoretical phase-difference estimation; 2) RS interferogram formation; and 3) comparison of theoretical and real phase difference distributions. The proposed method was validated on RS data acquired in Greenland and provides a proof of concept for the surface clutter discrimination using RS data.

In this paper, a segmentation-based approach to fine registration of multispectral and multitemporal very high resolution (VHR) images is proposed. The proposed approach aims at estimating and correcting the residual local misalignment [also referred to as registration noise (RN)] that often affects multitemporal VHR images even after standard registration. The method extracts automatically a set of object representative points associated with regions with homogeneous spectral properties (i.e., objects in the scene). Such points result to be distributed all over the considered scene and account for the high spatial correlation of pixels in VHR images. Then, it estimates the amount and direction of residual local misalignment for each object representative point by exploiting residual local misalignment properties in a multiple displacement analysis framework. To this end, a multiscale differential analysis of the multispectral difference image is employed for modeling the statistical distribution of pixels affected by residual misalignment (i.e., RN pixels) and detect them. The RN is used to perform a segmentation-based fine registration based on both temporal and spatial correlation. Accordingly, the method is particularly suitable to be used for images with a large number of border regions like VHR images of urban scenes. Experimental results obtained on both simulated and real multitemporal VHR images confirm the effectiveness of the proposed method.

The knowledge of the tree species is a crucial information that governs the success of precision forest management practice. High-density small footprint multireturn airborne light detection and ranging (LiDAR) scanning can collect a huge amount of point samples containing structural details of the forest vertical profile, which can reveal important structural information of the forest components. LiDAR data have been successfully used to distinguish between coniferous and deciduous/broadleaved tree species. However, species classification within a class (e.g., the conifer class) using LiDAR data is a challenging problem when considering the tree external crown characteristics only. This paper presents a novel method for conifer species classification based on the use of geometric features describing both the internal and external structures of the crown. The internal crown geometric features (IGFs) are defined based on a novel internal branch structure model, which uses 3-D region growing and principal component analysis to delineate the branch structure of a conifer tree accurately. IGFs are used together with external crown geometric features to perform conifer species classification. Three different support vector machines have been considered for classification performance evaluation. The experimental analysis conducted on high-density LiDAR data acquired over a portion of the Trentino region in Italy proves the effectiveness of the proposed method.

Tree crown structural parameters are key inputs to studies spanning forest fire propagation, invasive species dynamics, avian habitat provision, and so on, but these parameters consistently are difficult to measure. While airborne laser scanning (ALS) provides uniform data and a consistent nadir perspective necessary for crown segmentation, the data characteristics of terrestrial laser scanning (TLS) make such crown segmentation efforts much more challenging. We present a data fusion approach to extract crown structure from TLS, by exploiting the complementary perspective of ALS. Multiple TLS point clouds are automatically registered to a single ALS point cloud by maximizing the normalized cross correlation between the global ALS canopy height model (CHM) and each of the local TLS CHMs through parameter optimization of a planar Euclidean transform. Per-tree canopy segmentation boundaries, which are reliably obtained from ALS, can then be adapted onto the more irregular TLS data. This is repeated for each TLS scan; the combined segmentation results from each registered TLS scan and the ALS data are fused into a single per-tree point cloud, from which canopy-level structural parameters readily can be extracted.

Automatic registration of multimodal remote sensing data [e.g., optical, light detection and ranging (LiDAR), and synthetic aperture radar (SAR)] is a challenging task due to the significant nonlinear radiometric differences between these data. To address this problem, this paper proposes a novel feature descriptor named the histogram of orientated phase congruency (HOPC), which is based on the structural properties of images. Furthermore, a similarity metric named HOPCncc is defined, which uses the normalized correlation coefficient (NCC) of the HOPC descriptors for multimodal registration. In the definition of the proposed similarity metric, we first extend the phase congruency model to generate its orientation representation and use the extended model to build HOPCncc. Then, a fast template matching scheme for this metric is designed to detect the control points between images. The proposed HOPCncc aims to capture the structural similarity between images and has been tested with a variety of optical, LiDAR, SAR, and map data. The results show that HOPCncc is robust against complex nonlinear radiometric differences and outperforms the state-of-the-art similarities metrics (i.e., NCC and mutual information) in matching performance. Moreover, a robust registration method is also proposed in this paper based on HOPCncc, which is evaluated using six pairs of multimodal remote sensing images. The experimental results demonstrate the effectiveness of the proposed method for multimodal image registration.

In hyperspectral image analysis, the classification task has generally been addressed jointly with dimensionality reduction due to both the high correlation between the spectral features and the noise present in spectral bands, which might significantly degrade classification performance. In supervised classification, limited training instances in proportion with the number of spectral features have negative impacts on the classification accuracy, which is known as Hughes effects or curse of dimensionality in the literature. In this paper, we focus on dimensionality reduction problem, and propose a novel feature-selection algorithm, which is based on the method called high dimensional model representation. The proposed algorithm is tested on some toy examples and hyperspectral datasets in comparison with conventional feature-selection algorithms in terms of classification accuracy, stability of the selected features and computational time. The results show that the proposed approach provides both high classification accuracy and robust features with a satisfactory computational time.

Variational models are known to work well for addressing image restoration/regularization problems. However, most of the methods proposed in the literature are defined for scalar inputs and are used on multiband images (such as RGB or multispectral imagery) by the composition of a simple band-wise processing. This involves suboptimal results and may introduce artifacts. Only in a few cases, variational models are extended to the case of vector-valued inputs. However, the known implementations are restricted to the first-order models, while the second-order models are never considered. Thus, typical problems of the first-order models, such as the staircasing effect cannot be overtaken. This paper considers a second-order functional model to function approximation with free discontinuities given by Blake-Zisserman (BZ) and proposes an efficient minimization algorithm in the case of vector-valued inputs. In the BZ model, the Hessian of the solution is penalized outside a set of finite length, therefore the solution is forced to be piecewise linear. Moreover, the model allows the formation of free discontinuities and free gradient discontinuities. The proposed algorithm is applied to difficult color image restoration/regularization problems and to piecewise linear approximation of curves in space.

This paper presents a novel quad-tree based compressed histogram attribute profile (QT-CHAP) for classification of very high resolution remote sensing images. The QT-CHAP characterizes the marginal local distribution of attribute filter responses to model the spatial context of each sample with a very small number of image features. This is achieved based on a three steps algorithm that comprises a novel non-uniform quantization strategy to the compression of the information present in standard histogram attribute profiles. Due to the proposed quad-tree based non-uniform quantization strategy, the proposed QT-CHAP results in an optimized tradeoff between information extraction and number of considered features. Experimental results confirm the effectiveness of the proposed QT-CHAP in terms of computational complexity, storage requirements and classification accuracy when compared to the other state of the art attribute profile based methods.

When dealing with optical images, the most common approach to unsupervised change detection is Change Vector Analysis (CVA) which computes the multispectral difference image and exploits its statistical distribution in (hyper-)spherical coordinates. The latter step usually requires assumptions on both the model of class distributions and the number of changes. However, both assumptions are seldom satisfied especially when multisensor VHR images are considered. Thus, we propose an approach to multiple change detection in multisensor VHR optical images based on iterative clustering in (hyper-) spherical coordinate. The proposed approach is distribution free, unsupervised and automatically identifies the number of changes. Results obtained on a multitemporal and multisensor dataset including images from WorldView-2 and QuickBird are promising.

The exploitation of the high revisit time (8-16 days) by the COSMO-SkyMed®(CSK®) satellites is an important opportunity for agricultural mapping. This study aims at evaluating CSK®potentiality to classify different crop types, with CSK®multi-temporal images collected over the agricultural site of Marchfeld, in Austria. Two different time series of CSK®HIMAGE SAR scenes, at 3m resolution, 9 at HH and 9 at VH polarization were taken during the vegetation season (from April to October 2014). CSK®data were processed and analyzed to investigate crop signatures from CSK®backscattering coefficient of five crop types, namely carrot, corn, potato, soybean and sugarbeet. In situ field observations were conducted during the SAR data acquisition. CKS®data were overlaid with crop fields ground truth. A Support Vector Machine (SVM) classification method has been applied. The classification results yield very promising overall classification accuracies using the combination of HH and VH polarization.

The knowledge about the species of trees is essential for precision forest management practices. Modern high density airborne Light Detection and Ranging (LiDAR) systems have the ability to acquire large number of LiDAR points, allowing a very detailed characterization of the forest at the individual tree level. In this context, it is possible to use LiDAR data for accurate classification of the tree species. In this paper, we consider the specific problem of species classification of trees belonging to the conifer class. This is particularly challenging when only the external geometric information is considered. To address the problem we propose a novel approach that model the internal crown structure of the conifers. The internal structure is identified by using 3D region growing and Principal Component Analysis (PCA) and is used for defining a set of novel Internal Crown Geometric features (IGFs). Some state-of-the-art External Crown Geometric Features (EGFs) were also used to improve the classification accuracy. Sparse Support Vector Machines (SSVM) was used for classification and to quantify the feature relevances.

Supervised classification of hyperspectral images is a challenging task due to the relatively low ratio between the number of training samples and the number of spectral channels. Subspace-based classification methods deal with this difficulty by assuming that feature vectors lie in a low-dimensional subspace. Based on the fact that a class in a hyperspectral image may be composed of a number of different groups of materials and mixture of spectral features, we suggest to estimate several lower dimensional random subspaces for the samples within each class. For subspace learning and classification, we propose to exploit the union of random subspaces in a Gaussian Mixture Model. Experimental results, conducted on two real hyperspectral data sets, indicate that the proposed method provides competitive classification results in comparison with other state-of-the-art approaches.

Light Detection And Ranging (LiDAR) data have proven to be very effective in the estimation of parameters for forestry applications. However, little research has been done regarding the multitemporal analysis of these data. In this paper we propose a novel hierarchical change detection approach that first performs the detection of major changes (e.g., harvested trees) and then focuses on the detection of minor changes (e.g., single tree growth), using multitemporal LiDAR data having different point densities. Splitting the change detection problem allows us to analyze the different types of changes with different techniques. In particular, the detection of minor changes is carried out directly on the point clouds in order to exploit all the informative content of the LiDAR data. The approach has been tested on a dataset acquired in 2010 and 2014 on a complex forest area located in the Southern Italian Alps. The experimental results confirm the effectiveness of the proposed approach.

This paper analyzes and compares different Multiple Kernel Learning (MKL) algorithms for the classification of remote sensing (RS) images. The main purpose of the comparison is to identify advantages and disadvantages of different MKL algorithms in terms of their computational time and classification accuracy. Furthermore, some guidelines on the proper selection of the MKL algorithms associated with different RS image classification problems are derived.

In this paper we present a growth-model based approach to the accurate estimation of stem diameter at single tree level by using high-density LiDAR data. First, we detect classes of trees characterized by different growth conditions by means of a data-driven inference process. To this end, all the environmental factors that can affect the growth of the tree (i.e., forest density and topography) are modeled and analyzed. Second, for each detected growth-model class a tailored regression function is trained to adapt the model on the considered class. The crown structure, the topography and the forest density are considered to accurately retrieve the stem diameter. Experiments carried out in mountainous scenario characterized by complex morphology and a wide range of soil fertility demonstrate the effectiveness of the proposed method.

Recently, a thresholding method based on the Rayleigh-Rice mixture has been proposed for solving binary change detection problems in multispectral image pairs. However, when images acquired by the last generation of multispectral scanners having high radiometric resolution are considered, the distribution fitting is still not satisfactory and computed thresholds remain quite distant from the optimal values. The main reason for this seems to be that in all previous approaches the unchange class is modeled as a single class. Instead, both practice and recent studies showed that this is not the case for new generation data. In this work, we propose a generalized statistical model for the difference image that allows the unchange class to be complex. The resulting model has more degrees of freedom, therefore it better fits real distributions and returns almost optimal thresholds for binary decision also with high radiometric resolution images.

Typical tiling approaches to segmentation of large images perform separated runs of a specific segmentation algorithm on tiles and then merge the results. However, specific post-processing is often required to remove possible artifacts on tiles junctions. In this paper, we aim at showing that a simple tiling strategy with partially overlapping tiles can be applied to a 2-nd order variational segmentation method based on the minimization of the Blake-Zisserman functional, in such a way that tile boundaries are coherent without any need of specific post-processing. Moreover, the energy minimization is performed on each tile with Dirichlet initial boundary conditions; thus, tiles are independent and the whole procedure is parallelizable with independent tiles.

This letter introduces a novel unsupervised graph-theoretic approach in the framework of region-based retrieval of remote sensing (RS) images. The proposed approach is characterized by two main steps: (1) modeling each image by a graph, which provides region-based image representation combining both local information and related spatial organization, and (2) retrieving the images in the archive that are most similar to the query image by evaluating graph-based similarities. In the first step, each image is initially segmented into distinct regions and then modeled by an attributed relational graph, where nodes and edges represent region characteristics and their spatial relationships, respectively. In the second step, a novel inexact graph matching strategy, which jointly exploits a subgraph isomorphism algorithm and a spectral graph embedding technique, is applied to match corresponding graphs and to retrieve images in the order of graph similarity. Experiments carried out on an archive of aerial images point out that the proposed approach significantly improves the retrieval performance compared to the state-of-the-art unsupervised RS image retrieval methods.(RS) images.

Even after coregistration, very high resolution (VHR) multitemporal images acquired by different multispectral sensors (e.g., QuickBird and WordView) show a residual misregistration due to dissimilarities in acquisition conditions and in sensor properties. Residual misregistration can be considered as a source of noise and is referred to as registration noise (RN). Since RN is likely to have a negative impact on multitemporal information extraction, detecting and reducing it can increase multitemporal image processing accuracy. In this letter, we propose an approach to identify RN between VHR multitemporal and multisensor images. Under the assumption that dominant RN mainly exists along boundaries of objects, we propose to use edge information in high frequency regions to estimate it. This choice makes RN detection less dependent on radiometric differences and thus more effective in VHR multisensor image processing. In order to validate the effectiveness of the proposed approach, multitemporal multisensor data sets are built including QuickBird and WorldView VHR images. Both qualitative and quantitative assessments demonstrate the effectiveness of the proposed RN identification approach compared to the state-of-the-art one.

This paper presents spiking neural networks (SNNs) for remote sensing spatiotemporal analysis of image time series, which make use of the highly parallel and low-power-consuming neuromorphic hardware platforms possible. This paper illustrates this concept with the introduction of the first SNN computational model for crop yield estimation from normalized difference vegetation index image time series. It presents the development and testing of a methodological framework which utilizes the spatial accumulation of time series of Moderate Resolution Imaging Spectroradiometer 250-m resolution data and historical crop yield data to train an SNN to make timely prediction of crop yield. The research work also includes an analysis on the optimum number of features needed to optimize the results from our experimental data set. The proposed approach was applied to estimate the winter wheat (Triticum aestivum L.) yield in Shandong province, one of the main winter-wheat-growing regions of China. Our method was able to predict the yield around six weeks before harvest with a very high accuracy. Our methodology provided an average accuracy of 95.64%, with an average error of prediction of 0.236 t/ha and correlation coefficient of 0.801 based on a nine-feature model.

This paper presents a novel hybrid method to the estimation of bio/geophysical parameters, which models and corrects deviations from correct target values when theoretical electromagnetic models are used for the inversion process. The proposed hybrid method integrates theoretical models with empirical observations associated to a few field reference samples. This is achieved based on two steps. In the first step, deviations between estimations obtained by a theoretical model and empirical observations are initially computed. Then, deviations associated to unlabeled samples (for which reference measures are not existing) are characterized based on two different strategies: 1) the global deviation bias strategy (which assumes that the deviations of samples are constant within the input space); and 2) the local deviation bias strategy (which assumes that the deviations of samples are variable within different portions of the input space). In the second step, the theoretical model estimates of unlabeled samples are corrected based on the estimated deviations. The experimental analysis carried out in the context of soil moisture content retrieval from microwave remotely sensed data confirms the effectiveness of the proposed hybrid estimation method.

Large-scale remote sensing (RS) image search and retrieval have recently attracted great attention, due to the rapid evolution of satellite systems, that results in a sharp growing of image archives. An exhaustive search through linear scan from such archives is time demanding and not scalable in operational applications. To overcome such a problem, this paper introduces hashing-based approximate nearest neighbor search for fast and accurate image search and retrieval in large RS data archives. The hashing aims at mapping high-dimensional image feature vectors into compact binary hash codes, which are indexed into a hash table that enables real-time search and accurate retrieval. Such binary hash codes can also significantly reduce the amount of memory required for storing the RS images in the auxiliary archives. In particular, in this paper, we introduce in RS two kernel-based nonlinear hashing methods. The first hashing method defines hash functions in the kernel space by using only unlabeled images, while the second method leverages on the semantic similarity extracted by annotated images to describe much distinctive hash functions in the kernel space. The effectiveness of considered hashing methods is analyzed in terms of RS image retrieval accuracy and retrieval time. Experiments carried out on an archive of aerial images point out that the presented hashing methods are much faster, while keeping a similar (or even higher) retrieval accuracy, than those typically used in RS, which exploit an exact nearest neighbor search.

Morphological attribute profiles (APs) obtained by the sequential application of morphological attribute filters to images have been found very effective in remote sensing (RS) to characterize spatial properties of objects in a scene. However, a direct use of the APs can be insufficient to provide a complete characterization of spatial information when complex texture is present in the considered images. To overcome this problem, in this paper, we present the novel histogram-based morphological APs (HAPs). The HAPs model the marginal local distribution of attribute filter responses to better characterize the texture information, and they are obtained based on a three-step algorithm. In the first step, the standard APs are constructed by sequentially applying attribute filters to the considered image. In the second step, a local histogram is calculated for each sample of each image in the APs. Then, in the final step, the local histograms of the same pixel locations in the APs are stacked, resulting in a texture descriptor whose components represent local distributions of the filter responses for the related pattern. Finally, the very-high-dimensional HAPs are classified by a support vector machine (SVM) classifier with histogram intersection kernel. Experimental results obtained by considering two very high resolution panchromatic images show the effectiveness of the proposed HAPs, which sharply improve the accuracy of the SVM classifier with respect to standard AP-based methods.

This paper presents a novel multitemporal spectral unmixing (MSU) approach to address the challenging multiple-change detection problem in bitemporal hyperspectral (HS) images. Differently from the state-of-the-art methods that are mainly designed at a pixel level, the proposed technique investigates the spectral-temporal variations at a subpixel level. The considered change detection (CD) problem is analyzed in a multitemporal domain, where a bitemporal spectral mixture model is defined to analyze the spectral composition within a pixel. Distinct multitemporal endmembers (MT-EMs) are extracted according to an automatic and unsupervised technique. Then, a change analysis strategy is designed to distinguish the change and no-change MT-EMs. An endmember-grouping scheme is applied to the changed MT-EMs to detect the unique change classes. Finally, the considered multiple-change detection problem is solved by analyzing the abundances of the change and no-change classes and their contribution to each pixel. The proposed approach has been validated on both simulated and real multitemporal HS data sets presenting multiple changes. Experimental results confirmed the effectiveness of the proposed method.

A novel set-to-set distance-based spectral-spatial classification method for hyperspectral images (HSIs) is proposed. In HSIs, the spatially connected and spectrally similar pixels within each homogeneous region can be considered as one set of test samples, i.e., a test set, which should belong to the same class. In addition, each class of labeled pixels can be regarded as one set of training samples, i.e., a training set. Therefore, it is a natural consideration in the proposed method to measure the similarity between test and training sets via specific set-based distance criteria and then decide the classification label for each test set, accordingly. Specifically, the superpixel algorithm-based oversegmentation technique jointly exploits both the spatial similarity and structural information to first divide the HSI into multiple perceptually uniform regions. As a result, each segmented region corresponds to one test set. Then, each test/training set is represented with an affine hull (AH) model, which utilizes both the similarity and variance of pixels within each set to adaptively characterize the set. Finally, the class label for each test set is determined based on the closest geometry distance between test and training AHs. Experimental results on real HSI data sets demonstrate the superiority of the proposed algorithm over several well-known classification approaches, in terms of classification accuracy and computational speed.

Small-footprint high-density LiDAR data provide information on both the dominant and the subdominant layers of the forest. However, tree detection is usually carried out in the Canopy Height Model (CHM) image domain, where not all the dominant trees are distinguishable and the understory vegetation is not visible. To address these issues, we propose a novel method that integrates the analysis of the CHM with that of the point cloud space (PCS) to 1) improve the accuracy in the detection and delineation of the dominant trees and 2) identify and delineate the subdominant trees. By means of a derivative analysis of the horizontal profile of the forest, the method detects the missed crowns and delineates the crown boundaries directly in the PCS. Then, for each segmented crown, the vertical profile is analyzed to identify the presence of subcanopies and extract them. The proposed method does not require any prior knowledge on the stand properties (e.g., crown size and forest density). Experimental results obtained on two LiDAR data sets characterized by different laser point density show that the proposed method always improved the detection rate compared to other state-of-the-art techniques. It correctly detected 97% and 92% of the dominant trees measured in situ in high- and low-density LiDAR data, respectively. Moreover, it automatically identified 77% of the subdominant trees manually extracted by an expert operator in the high-density LiDAR data.

This paper presents a kernel-based feature selection method for the classification of hyperspectral images. The proposed method aims at selecting a subset of the original features that are both 1) relevant (discriminant) for the considered classification problem, i.e., preserve the functional relationship between input and output variables, and 2) invariant (stable) across different domains, i.e., minimize the data-set shift between the source and the target domains. Domains can be associated with hyperspectral images collected either on different geographical areas or on the same area at different times. We propose a novel measure of data-set shift for evaluating the domain stability, which computes the distance of the conditional distributions between the source and target domains in a reproducing kernel Hilbert space. Such a measure is defined on the basis of the kernel embeddings of the conditional distributions resulting in a nonparametric approach that does not require estimating the distribution of the classes. The adopted search strategy is based on a multiobjective optimization algorithm, which optimizes the two terms of the criterion function for the estimation of the Pareto-optimal solutions. This results in an effective approach of performing feature selection in a transfer learning setting. The experimental results obtained on two hyperspectral images show the effectiveness of the proposed method in selecting features with high generalization capabilities.

Wireless sensor networks offer unprecedented opportunities to monitor natural ecosystems. However, despite the growing number of applications (e.g., Forest fire detection, wildlife monitoring), the deployment challenges posed by the real-world natural environment still hinder the widespread adoption of this technology. In particular, the unpredictability of the low-power wireless channel in the presence of vegetation requires costly trial-and-error pilot campaigns to understand where and how to place the wireless nodes. In this paper, we propose a technique based on remote sensing for accurately estimating low-power radio signal attenuation in forest environments. We leverage airborne Light Detection and Ranging (LiDAR) instruments and related automatic data analysis systems to determine local forest attributes (e.g., Tree density) that, once factored into a specialized radio path loss model, enable accurate estimation of the received signal power. Our approach is i) automatic, i.e., It does not require in-field campaigns, and ii) fine-grained, i.e., It enables per-link estimates. Our validation from deployments in a real forest shows that the error of our per-link estimates of the received signal power is around ± 6 dBm - the accuracy of RSSI readings from the radio transceiver.

Given a user-defined image time series (i.e., the query time series), content based time series retrieval (CBTSR) is the process of identifying other time series that show properties similar to the query. When dealing with time series, the elements of the content based retrieval process require to be redefined in order to take into account the time variable. In this perspective, the design of the query, the feature extraction, and retrieval itself have to be reformulated. Here we focus our attention to CBTSR in pairs of images. The goal is to identify bi-temporal images showing a specific kind of change (associated with changes on the ground) modeled by the query. Attention is devoted to the design of the auxiliary archive modeling the change information and on the retrieval algorithm. Experiments on an archive of Landsat images confirmed the effectiveness of the proposed approach.

This paper presents the Radar for Icy Moon Exploration (RIME) that is a fundamental payload in the Jupiter Icy Moon Explorer (JUICE) mission of the European Space Agency (ESA). RIME is a radar sounder aimed to study the subsurface of Jupiter's icy moons Ganymede, Europa and Callisto. The paper illustrates the main goals of RIME, its architecture and parameters and some recent advances in its design.

In spaceborne radar sounder systems, the constraint on data storage imposes an upper bound to the radar receiving window length. Accordingly, an adaptive tracking procedure capable of shifting the receiving window in time according to the topography and spacecraft orbit variations is needed for maximizing the amount of signal acquired from the subsurface. Moreover, spaceborne sounder systems usually have constraints in terms of number of acquisitions and power consumption. Thus, tracking robustness and simplicity are mandatory requirements. In this paper, we present an adaptive range tracking technique derived from a combination of OCOG (Offset Center Of Gravity) leading edge estimation and a-p1filter. It gives significant improvements in the loss of lock detection. The performances of the overall technique are evaluated by different simulations of meaningful application scenarios which confirm its effectiveness.

Surface clutter can corrupt both Earth and planetary Radar Sounder (RS) observations preventing definitive interpretation of subsurface features, which are often the primary interest of geologists and planetary scientists. Clutter is usually detected by manual or automatic techniques that require ancillary information about the topography of the surface. However, this topography information is not always available. In this paper, we propose a novel method for clutter detection that is independent from ancillary information. This method uses a two channel RS system to exploit the cross-channel interferometric phase difference and is made up three main steps: i) feature extraction and theoretical phase difference estimation, ii) RS interferogram formation and iii) comparison of theoretical and real phase difference distributions. The proposed method has been validated on RS data acquired in Greenland.

The availability of multitemporal images acquired by several very high geometrical resolution (VHR) optical sensors makes it possible to build VHR image Time-Series (TS) with a temporal resolution better than the one achievable when considering a single sensor. However, such TS include images showing different characteristics from the geometrical, radiometrical and spectral viewpoint. Thus, there is a need of methods for building consistent VHR optical TS when using multispectral Multi-Sensor (MS) images. Here we focus on the spectral domain only, by designing a method to transform one image in an MS-TS into the spectral domain of another image in the same MS-TS, but acquired by a different sensor. To this end, a prediction-based approach relying on Artificial Neural Networks (ANN) is employed. In order to mitigate the impacts of possible changes occurred on the ground, the prediction model estimation is based on unchanged samples only. Experimental results obtained on VHR optical MS images confirm the effectiveness of the proposed approach.

Kernel-based image classification methods rely on the considered kernel functions that can be chosen with respect to prior information on the adopted features. In remote sensing, histogram features have recently gained an increasing interest due to their capability to address several critical classification problems (e.g., the problem of curse of dimensionality) when appropriate kernels and classifiers are selected. In view of that, in this paper we introduce in remote sensing additive kernels in the context of support vector machine classification (AK-SVM), which are suitable kernels for histogram based feature representations. In particular, we investigate the Histogram Intersection kernel and the chi-square kernel within the AK-SVM. Moreover, we present fast implementations of the AK-SVM to significantly speed up the classification phase of the SVM. Experimental results show the effectiveness of the AK-SVM in terms of classification accuracy and computational time when compared to SVMs with standard kernels.

This paper presents a novel histogram based attribute profiles (HAPs) technique for classification of very high resolution remote sensing images. The HAPs characterize the marginal local distribution of attribute filter responses to model the texture information. This is achieved based on a two steps algorithm. In the first step the standard attribute profiles (AP) are built through sequential application of attribute filters to the considered image. In the second step a local histogram is initially computed for each sample of each image in the APs. Then the local histograms of the same pixel locations in the APs are concatenated. Accordingly, each sample is characterized by a texture descriptor whose components model local distributions of the filter responses. Finally the very high dimensional HAPs are classified by a Support Vector Machine classifier with histogram intersection kernel, which is very effective for high dimensional histogram-based feature representations. Experimental results confirm the effectiveness of the proposed HAPs with respect to standard APs.

Very High Resolution (VHR) multitemporal images show a residual misalignment even after applying effective state of the art co-registration. This residual misalignment is caused by the dissimilarities of the acquisition circumstances such as off-nadir angle of the sensor, stability of the acquisition platform, structure of the considered scene, and so on. This paper aims at mitigating the residual misalignment of VHR multitemporal images to get a fine co-registration result. Here we propose to use Registration Noise (RN), which represents misaligned samples, for refining co-registration. After standard co-registration, a local analysis of RN pixels is fulfilled for extracting Control Points (CPs) and matching them according to the amount of the RN pixels. Matched CPs are employed for generating a deformation map to warp one image to the other image. Experiments carried out on both simulated and real multitemporal VHR images acquired by QuickBird sensors confirm the validity of the analysis and effectiveness of the proposed method.

The rise in global temperature accelerates the melting of the ice sheets. This highlights the importance of understanding the ice sheet structure and dynamics. Although fundamental for a reliable 3D modeling of the ice subsurface, the development of automatic techniques for the integration of multisensor data acquired over the ice sheets is still very limited. To address this issue, in this paper we present an automatic technique that integrates two types of data, i.e., radargrams (vertical profiles of the ice) acquired by airborne radar sounder (RS) instruments and ice surface elevation data recorded by altimeter (ALT) instruments. The aim of the technique is to perform a 3D reconstruction of the ice sheet by estimating maps of the surface, bedrock and ice thickness at the best possible reliable scale. Initially, the estimations of both the surface map and the related uncertainty intervals are done at several different scales by geostatistically interpolating information extracted from the RS data. Afterwards, a validation of the interpolation results and the identification of the best interpolation scale are performed by integrating the ALT data. The best scale is then used to estimate the bedrock elevation map, following the same interpolation scheme used for the estimation of the surface elevation map. Finally, the ice thickness map is computed by subtraction. The method has been applied to a subset of RS and ALT data acquired over a portion of the Byrd Glacier in Antarctica.

In hyperspectral image analysis, the classification task has generally been discussed with dimensionality reduction due to high correlation and noise between the spectral features, which might cause significantly low classification performance. In supervised classification, limited training samples in proportion to the number of spectral features have also negative impacts on the classification accuracy, which has known as Hughes effects or curse of dimensionality in the literature. In this paper, we focus on dimensionality reduction problem, and proposed a novel feature selection algorithm by using the method called random sampling high dimensional model representation (RS-HDMR), and the proposed algorithm were tested on a toy and hyperspectral dataset in comparison to conventional feature selection algorithms with regards to both computational time and classification accuracy.

Hyperspectral data classification problems have been extensively studied in the past decade. However, well designed features and a robust classifier are still open issues that impact on the performance of an automatic land-cover classification system. In this paper, we propose a deep feature represenation method that generates very good features and a classifier for pixel-wise hyperspectral data classification. The proposed method has two main steps: principle components of the hyperspectral image cube is first filtered by three dimensional Gabor wavelets; second, stacked autoencoders are trained on the outputs of the previous step through unsupervised pre-training, finally deep neural network is trained on those stacked autoencoders. Experimental results obtained on real hyperspectral image confirmed the effectiveness of the proposed approach in favors of the high classification accuracy and computation efficiency.

This paper develops a novel multitemporal spectral unmixing (MSU) approach for addressing the challenging multiple-change detection problem in bi-temporal hyperspectral (HS) images. Differently from state-of-the-art techniques that mainly perform at a pixel level, the proposed MSU approach investigates the spectral-temporal variations at a subpixel level. A multitemporal spectral mixture model is defined to analyze the spectral composition within a pixel. Distinct multitemporal endmembers (MT-EMs) are extracted and employed for distinguishing change and no-change MT-EMs in the unmixing model. The CD problem is solved by analyzing the abundances of the unique change and no-change multitemporal endmembers and their contribution to each pixel. Experimental results obtained on multitemporal Hyperion HS images confirmed the effectiveness of the proposed method.

This research aims at analyzing the integration of C and X band data collected from Radarsat2 (RS2) and COSMO-SkyMed (CSK) systems on some test areas in Italy, in order to estimate the main geophysical parameters of soil and vegetation, such as soil moisture and vegetation biomass. A check of the sensitivity of SAR signal to the soil parameters was first carried out on both test sites. Over the South-Tyrol area a retrieval approach based on the Support Vector Regression methodology, which was already tested in this area using C-band data from ENVISAT/ASAR data, was carried out. From these preliminary results it can be concluded that X-band images combined with C-band images could provide valuable information for the retrieval of SMC, even though further investigations should be carried out on a larger time-series and larger set of samples.

Modern forest inventory is based on the accurate and precise characterization of the 3D structure of the forest. Although LiDAR (Light Detection and Ranging) is an effective tool to estimate forest parameters, when acquired from single view point it is not able to represent accurately the entire scene. To solve this problem, in this paper we present a method that integrates the terrestrial and airborne LiDAR data. The proposed method first performs an automatic co-registration of the data sources based on the spatial pattern of the structure of the stand plot. Second, it integrates the LiDAR point clouds to accurately represent the structure of the crown. The resulting fused LiDAR point cloud can be used for an accurate estimation of the crown parameters, thus making it possible a more comprehensive representation of the 3D structure of the forest stand. Experimental results carried out in a oakland savanna in Fresno (California) confirm the effectiveness of the proposed method.

In this paper we present a hierarchical approach to the segmentation of high-density LiDAR data which aims to automatically detect and delineate the single tree crowns of both the dominant and the dominated layers of the forest. First, we detect the dominant tree crowns by using both the image derived from the LiDAR data and the LiDAR point cloud. Hence, the detected crowns are delineated directly in the LiDAR point cloud by means of a radial angular analysis. Second, the dominated crowns are detected by analyzing the vertical profile of the dominant trees. Finally, we extract the dominated trees, thus reconstructing the structure of the forest. Experiments carried out in a forest area located in the Southern Italian Alps by using very high density LiDAR data (up to 50 points/m2) point out the effectiveness of the proposed approach.

Rough set theory is a paradigm to deal with uncertainty, vagueness, and incompleteness of data. Although it has been applied successfully to feature selection in different application domains, it is seldom used for the analysis of hyperspectral images. In this paper, a rough set based supervised method is proposed to select informative bands in hyperspectral images. The proposed technique exploits rough set theory to define a novel criterion for selecting informative bands. The performances of the proposed approach were compared with those of three state-of-the-art methods on a hyperspectral data set. Experimental results show the effectiveness of the proposed technique.

This letter addresses the problem of unsupervised land-cover classification of remotely sensed multispectral satellite images from the perspective of cluster ensembles and self-learning. The cluster ensembles combine multiple data partitions generated by different clustering algorithms into a single robust solution. A cluster-ensemble-based method is proposed here for the initialization of the unsupervised iterative expectation-maximization (EM) algorithm which eventually produces a better approximation of the cluster parameters considering a certain statistical model is followed to fit the data. The method assumes that the number of land-cover classes is known. A novel method for generating a consistent labeling scheme for each clustering of the consensus is introduced for cluster ensembles. A maximum likelihood classifier is henceforth trained on the updated parameter set obtained from the EM step and is further used to classify the rest of the image pixels. The self-learning classifier, although trained without any external supervision, reduces the effect of data overlapping from different clusters which otherwise a single clustering algorithm fails to identify. The clustering performance of the proposed method on a medium resolution and a very high spatial resolution image have effectively outperformed the results of the individual clustering of the ensemble.

In this letter, we explore the use of self-dual attribute profiles (SDAPs) for the classification of hyperspectral images. The hyperspectral data are reduced into a set of components by nonparametric weighted feature extraction (NWFE), and a morphological processing is then performed by the SDAPs separately on each of the extracted components. Since the spatial information extracted by SDAPs results in a high number of features, the NWFE is applied a second time in order to extract a fixed number of features, which are finally classified. The experiments are carried out on two hyperspectral images, and the support vector machines and random forest are used as classifiers. The effectiveness of SDAPs is assessed by comparing its results against those obtained by an approach based on extended APs.

This paper addresses the problem of land-cover classification of remotely sensed image pairs in the context of domain adaptation. The primary assumption of the proposed method is that the training data are available only for one of the images (source domain), whereas for the other image (target domain), no labeled data are available. No assumption is made here on the number and the statistical properties of the land-cover classes that, in turn, may vary from one domain to the other. The only constraint is that at least one land-cover class is shared by the two domains. Under these assumptions, a novel graph theoretic cross-domain cluster mapping algorithm is proposed to detect efficiently the set of land-cover classes which are common to both domains as well as the additional or missing classes in the target domain image. An interdomain graph is introduced, which contains all of the class information of both images, and subsequently, an efficient subgraph-matching algorithm is proposed to highlight the changes between them. The proposed cluster mapping algorithm initially clusters the target domain data into an optimal number of groups given the available source domain training samples. To this end, a method based on information theory and a kernel-based clustering algorithm is proposed. Considering the fact that the spectral signature of land-cover classes may overlap significantly, a postprocessing step is applied to refine the classification map produced by the clustering algorithm. Two multispectral data sets with medium and very high geometrical resolution and one hyperspectral data set are considered to evaluate the robustness of the proposed technique. Two of the data sets consist of multitemporal image pairs, while the remaining one contains images of spatially disjoint geographical areas. The experiments confirm the effectiveness of the proposed framework in different complex scenarios.

Conventional relevance feedback (RF) schemes improve the performance of content-based image retrieval (CBIR) requiring the user to annotate a large number of images. To reduce the labeling effort of the user, this paper presents a novel active learning (AL) method to drive RF for retrieving remote sensing images from large archives in the framework of the support vector machine classifier. The proposed AL method is specifically designed for CBIR and defines an effective and as small as possible set of relevant and irrelevant images with regard to a general query image by jointly evaluating three criteria: uncertainty; diversity; and density of images in the archive. The uncertainty and diversity criteria aim at selecting the most informative images in the archive, whereas the density criterion goal is to choose the images that are representative of the underlying distribution of data in the archive. The proposed AL method assesses jointly the three criteria based on two successive steps. In the first step, the most uncertain (i.e., ambiguous) images are selected from the archive on the basis of the margin sampling strategy. In the second step, the images that are both diverse (i.e., distant) to each other and associated to the high-density regions of the image feature space in the archive are chosen from the most uncertain images. This step is achieved by a novel clustering-based strategy. The proposed AL method for driving the RF contributes to mitigate problems of unbalanced and biased set of relevant and irrelevant images. Experimental results show the effectiveness of the proposed AL method.

The availability of hyperspectral images with improved spectral and spatial resolutions provides the opportunity to obtain accurate land-cover classification. In this paper, a novel methodology that combines spectral and spatial information for supervised hyperspectral image classification is proposed. A feature reduction strategy based on independent component analysis is the main core of the spectral analysis, where the exploitation of prior information coupled to the evaluation of the reconstruction error assures the identification of the best class-informative subset of independent components. Reduced attribute profiles (APs), which are designed to address well-known issues related to information redundancy that affect the common morphological APs, are then employed for the modeling and fusion of the contextual information. Four real hyperspectral data sets, which are characterized by different spectral and spatial resolutions with a variety of scene typologies (urban, agriculture areas), have been used for assessing the accuracy and generalization capabilities of the proposed methodology. The obtained results demonstrate the classification effectiveness of the proposed approach in all different scene typologies, with respect to other state-of-the-art techniques.

Even after applying effective coregistration methods, multitemporal images are likely to show a residual misalignment, which is referred to as registration noise (RN). This is because coregistration methods from the literature cannot fully handle the local dissimilarities induced by differences in the acquisition conditions (e.g., the stability of the acquisition platform, the off-nadir angle of the sensor, the structure of the considered scene, etc.). This paper addresses the problem of reducing such a residual misalignment by proposing a fine automatic coregistration approach for very high resolution (VHR) multispectral images. The proposed method takes advantage of the properties of the residual misalignment itself. To this end, RN is first extracted in the change vector analysis (CVA) polar domain according to the behaviors of the specific multitemporal images considered. Then, a local analysis of RN pixels (i.e., those showing residual misalignment) is conducted for automatically extracting control points (CPs) and matching them according to their estimated displacement. Matched CPs are used for generating a deformation map by interpolation. Finally, one VHR image is warped to the coordinates of the other through a deformation map. Experiments carried out on simulated and real multitemporal VHR images confirm the effectiveness of the proposed approach.

Exhaustive investigations of the ice sheet subsurface can be carried out by analyzing the information contained in the huge archives of radargrams acquired by dedicated radar sounder (RS) instruments. The analysis can be done by using properly designed automatic techniques for a quantitative, objective, and reliable extraction of information from radargrams. Unfortunately, the definition and development of such automatic techniques have only been marginally addressed in the literature. In this paper, we propose a novel and efficient system for the automatic classification of ice subsurface targets present in radargrams. The core of the system is represented by the extraction of a set of features for target discrimination. The features are based on both the specific statistical properties of the RS signal and the spatial distribution of the ice subsurface targets. Such features are then provided as input to an automatic classifier based on support vector machine. Experimental results obtained on two real-world data sets acquired by airborne-mounted RSs in large regions of Antarctica confirm the robustness and effectiveness of the proposed classification system.

The new generation of satellite hyperspectral (HS) sensors can acquire very detailed spectral information directly related to land surface materials. Thus, when multitemporal images are considered, they allow us to detect many potential changes in land covers. This paper addresses the change-detection (CD) problem in multitemporal HS remote sensing images, analyzing the complexity of this task. A novel hierarchical CD approach is proposed, which is aimed at identifying all the possible change classes present between the considered images. In greater detail, in order to formalize the CD problem in HS images, an analysis of the concept of “change” is given from the perspective of pixel spectral behaviors. The proposed novel hierarchical scheme is developed by considering spectral change information to identify the change classes having discriminable spectral behaviors. Due to the fact that, in real applications, reference samples are often not available, the proposed approach is designed in an unsupervised way. Experimental results obtained on both simulated and real multitemporal HS images demonstrate the effectiveness of the proposed CD method.

This paper presents an effective semiautomatic method for discovering and detecting multiple changes (i.e., different kinds of changes) in multitemporal hyperspectral (HS) images. Differently from the state-of-the-art techniques, the proposed method is designed to be sensitive to the small spectral variations that can be identified in HS images but usually are not detectable in multispectral images. The method is based on the proposed sequential spectral change vector analysis, which exploits an iterative hierarchical scheme that at each iteration discovers and identifies a subset of changes. The approach is interactive and semiautomatic and allows one to study in detail the structure of changes hidden in the variations of the spectral signatures according to a top-down procedure. A novel 2-D adaptive spectral change vector representation (ASCVR) is proposed to visualize the changes. At each level this representation is optimized by an automatic definition of a reference vector that emphasizes the discrimination of changes. Finally, an interactive manual change identification is applied for extracting changes in the ASCVR domain. The proposed approach has been tested on three hyperspectral data sets, including both simulated and real multitemporal images showing multiple-change detection problems. Experimental results confirmed the effectiveness of the proposed method.

The increasing availability of very high resolution (VHR) images regularly acquired over urban areas opens new attractive opportunities for monitoring human settlements at the level of individual buildings. This paper presents a novel approach to building change detection in multitemporal VHR synthetic aperture radar (SAR) images. The proposed approach is based on two concepts: 1) the extraction of information on changes associated with increase and decrease of backscattering at the optimal building scale and 2) the exploitation of the expected backscattering properties of buildings to detect either new or fully demolished buildings. Each detected change is associated with a grade of reliability. The approach is validated on the following: 1) COSMO-SkyMed multitemporal spotlight images acquired in 2009 on the city of L'Aquila (Italy) before and after the earthquake that hit the region and 2) TerraSAR-X multitemporal spotlight images acquired on the urban area of the city of Trento (Italy). Results demonstrate that the proposed approach allows an accurate identification of new and demolished buildings while presents a low false-alarm rate and a high reliability.

In this paper, we study the problem of feature extraction for knowledge transfer between multiple remotely sensed images in the context of land-cover classification. Several factors such as illumination, atmospheric, and ground conditions cause radiometric differences between images of similar scenes acquired on different geographical areas or over the same scene but at different time instants. Accordingly, a change in the probability distributions of the classes is observed. The purpose of this work is to statistically align in the feature space an image of interest that still has to be classified (the target image) to another image whose ground truth is already available (the source image). Following a specifically designed feature extraction step applied to both images, we show that classifiers trained on the source image can successfully predict the classes of the target image despite the shift that has occurred. In this context, we analyze a recently proposed domain adaptation method aiming at reducing the distance between domains, Transfer Component Analysis, and assess the potential of its unsupervised and semisupervised implementations. In particular, with a dedicated study of its key additional objectives, namely the alignment of the projection with the labels and the preservation of the local data structures, we demonstrate the advantages of Semisupervised Transfer Component Analysis. We compare this approach with other both linear and kernel-based feature extraction techniques. Experiments on multi- and hyperspectral acquisitions show remarkable cross- image classification performances for the considered strategy, thus confirming its suitability when applied to remotely sensed images.

Light detection and ranging (LiDAR) technology has been extensively used for estimating forest attributes. Although high-spatial-density LiDAR data can be used to accurately derive attributes at single tree level, low-density LiDAR data are usually acquired for reducing the cost. However, a low density strongly affects the estimation accuracy due to the underestimation of the tree top and the possible loss of crowns that are not hit by any LiDAR point. In this paper, we propose a 3-D model-based approach to the estimation of the tree top height based on the fusion between low-density LiDAR data and high-resolution optical images. In the proposed approach, the integration of the two remotely sensed data sources is first exploited to accurately detect and delineate the single tree crowns. Then, the LiDAR vertical measures are associated to those crowns hit by at least one LiDAR point and used together with the radius of the crown and the tree apex location derived from the optical image for reconstructing the tree top height by a properly defined parametric model. For the remaining crowns detected only in the optical image, we reconstruct the tree top height by proposing a k-nearest neighbor trees technique that estimates the height of the missed trees as the average of the k reconstructed height values of the trees having most similar crown properties. The proposed technique has been tested on a coniferous forest located in the Italian Alps. The experimental results confirmed the effectiveness of the proposed method.

Band selection is a well-known approach to reduce the dimensionality of hyperspectral imagery. Rough set theory is a paradigm to deal with uncertainty, vagueness, and incompleteness of data. Although it has been applied successfully to feature selection in different application domains, it is seldom used for the analysis of the hyperspectral imagery. In this paper, a rough-set-based supervised method is proposed to select informative bands from hyperspectral imagery. The proposed technique exploits rough set theory to compute the relevance and significance of each spectral band. Then, by defining a novel criterion, it selects the informative bands that have higher relevance and significance values. To assess the effectiveness of the proposed band selection technique, three state-of-the-art methods (one supervised and two unsupervised) used in the remote sensing literature are analyzed for comparison on three hyperspectral data sets. The results of this comparison point to the superiority of the proposed technique, especially when a small number of bands are to be selected.

In this paper, we present a novel domain adaptation technique aimed at providing reliable change detection maps for a series of image pairs acquired on the same area at different times. The proposed technique exploits the polar change vector analysis method and assumes that the reference data for characterizing a specific change of interest are available only for a pair of images (source domain). Then, it exploits the knowledge learned from the source domain and adapts it to other pairs of images belonging to the time series (target domains) to be analyzed. The proposed technique is able to handle possible radiometric differences among images adapting in an unsupervised way the decision rule estimated on the source domain to the target domains through variables estimated directly on the target images. The proposed approach has been applied to two data sets made up of time series of Landsat Thematic Mapper images. In one case, the change of interest is related to evolution of deforestation, while in the other case, it is related to burned area detection. Experimental results show the effectiveness of the proposed technique.

The problem of estimating the parameters of a Rayleigh-Rice mixture density is often encountered in image analysis (e.g., remote sensing and medical image processing). In this paper, we address this general problem in the framework of change detection (CD) in multitemporal and multispectral images. One widely used approach to CD in multispectral images is based on the change vector analysis. Here, the distribution of the magnitude of the difference image can be theoretically modeled by a Rayleigh-Rice mixture density. However, given the complexity of this model, in applications, a Gaussian-mixture approximation is often considered, which may affect the CD results. In this paper, we present a novel technique for parameter estimation of the Rayleigh-Rice density that is based on a specific definition of the expectation-maximization algorithm. The proposed technique, which is characterized by good theoretical properties, iteratively updates the parameters and does not depend on specific optimization routines. Several numerical experiments on synthetic data demonstrate the effectiveness of the method, which is general and can be applied to any image processing problem involving the Rayleigh-Rice mixture density. In the CD context, the Rayleigh-Rice model (which is theoretically derived) outperforms other empirical models. Experiments on real multitemporal and multispectral remote sensing images confirm the validity of the model by returning significantly higher CD accuracies than those obtained by using the state-of-the-art approaches.

This paper presents a novel approach to binary change detection in pairs of images extracted from time series. The main idea is that, given a binary change detection map obtained with any literature technique applied to the considered pair of images, we can identify possible change detection errors exploiting other images in the time series. This can be done by considering other pairs of images in the time series that, jointly with the analyzed one, can define a closed circular path in time. Then we model the binary change variable as a conservative field along circular paths within the time series. If for a pixel the circular path does not satisfy the conservativeness property an error is detected. Accordingly, the change detection label on that pixel is considered unreliable Experimental results obatined on a time series of ASAR Envisat images point out the effectiveness of the approach in detecting unreliable pixels.

In this paper we compare features obtained by different filtering strategies for morphological attribute filters by considering non-increasing attributes. The Attribute profiles (APs) and Self Dual Attribute Profiles (SDAPs) are obtained by sequentially applying attribute filters on tree-based image representations, such as Min- or Max-trees and Inclusion tree, respectively. This work aims to study the effects of using the filtering rules max, min, direct and subtractive, when considering the non-increasing attributes moment of inertia and standard deviation. A very high spatial resolution data set is used in the experiments, and the extracted information obtained by the profiles is analyzed. This is done by studying the effects on the classification accuracy by using the profiles as additional input features to a Random Forest classifier.

This paper presents hashing based approximate nearest neighbor search algorithms that allow fast and accurate image retrieval in huge remote sensing data archives. Hashing methods aim at mapping high-dimensional image feature vectors into short binary codes based on hashing functions. Then, the image retrieval is accomplished according to Hamming distances of image hash codes. In particular, in this paper two hashing methods are adopted for RS image retrieval problems. The former aims at defining hash functions in the kernel space by using only unlabeled images. The latter leverages on the semantic similarity given in terms of annotated images to define much distinctive hash functions in the kernel space. The effectiveness of both methods is analyzed in terms of RS image retrieval accuracy as well as retrieval time. Experiments carried out on an archive of aerial images show that the presented hashing methods are one hundred times faster than those that exploit an exact nearest neighbor search while keeping a high retrieval accuracy.

This article proposes a feature reduction technique for hyperspec-tral images using Independent Component Analysis (ICA). The proposed technique aims at extracting the best subset of class-informative independent components (ICs) for hyperspectral supervised classification. The selection of the most representative components is assured by the minimization of the reconstruction error, which is computed on the training samples used for the supervised classification. The searching strategy is optimized by exploiting a genetic algorithm-based approach where the fitness function is the classification accuracy obtained by using a support vector machine (SVM) classifier. The obtained results show the effectiveness of the proposed approach in providing class-informative components to improve the classification accuracy.

This paper focuses on a challenging task for representing and detecting multiple changes in multitemporal hyperspectral images. To this aim, a novel Sequential Spectral Change Vector Analysis (S2CVA) method is proposed that extends the use of the popular C2VA method [1]. The proposed S2CVA approach is designed in a sequential and semiautomatic fashion, where a fully automatic 2-D change representation and an interactive change identification are included at each level of the processing, exploiting the multiple change information hierarchically. In particular, an adaptive reference vector scheme is developed to drive the change representation, and thus the sequential analysis, by following a top-down structure. Changes are represented and separated according to their spectral change significance. Experimental results obtained on multitemporal Hyperion images confirm the effectiveness of the proposed method.

In this paper an approach that exploits and combines the acquisition modes offered by satellite SAR systems is presented that: i) quickly and automatically identifies the areas severely affected by a catastrophic event (i.e., hotspots), such as an earthquake by analyzing images characterized by a large coverage and a medium to high geometrical resolution; and ii) analyzes images characterized by very high geometrical resolution acquired over hot-spots in order to detect collapsed buildings. Experimental results obtained on a dataset made up of COSMO-SkyMed (CSK) data acquired before and after the 2009 L'Aquila earthquake (Italy) demonstrate the effectiveness of the proposed approach.

This paper proposes a novel neural model for unsupervised change detection in time series of multispectral remote sensing imagery using clustering with Self-Organizing Map (SOM) for automatic pseudo-training sample set selection cascaded with Concurrent Self-Organizing Maps (CSOM) classifier. The proposed algorithm has the following steps: (a) computation of difference image (DI) corresponding to the magnitudes of Spectral Change Vectors (SCVs); (b) SOM clustering to automatically deduce the SCV domain quantization parameters defining the pseudo-training sample set regions (changed, unchanged and uncertain); (c) CSOM classification. The model is evaluated using a Landsat-5 image set acquired on a Mexico area before and after two wildfires. As a benchmark, we have considered the classical method of Bayes theory-EM algorithm for selection of pseudo-training sample set combined with a S3VM classifier. The results confirm the effectiveness of our neural approach. Moreover, the exciting advantage of the proposed model over the classical ones is that it does not require any statistical assumptions regarding changed/unchanged SCVs data and it implies a reduced computational effort.

In mountain areas, soil moisture is a key parameter for both agricultural management and natural hazard support. This paper presents an approach for retrieval of soil moisture content (SMC) from different satellite sensors with a specific focus on mountain areas. The experimental analysis was carried out on images acquired over the Südtirol/Alto Adige Province (Italy) during 2010-2011 from the RADARSAT2 in quad-pol mode and Envisat ASAR in Wide Swath mode in VV polarization. The methodology for soil moisture retrieval is based on the Support Vector Regression (SVR) method specifically trained to be able to consider topographic effects of the mountain areas. The comparison with ground measurements collected during field campaigns indicates an RMSE value of around 5% of SMC% while the comparison with fixed ground stations reports an error of around 9% of SMC%. Comparing RADARSAT2 and ASAR SMC, both datasets reveal very similar distributions of SMC values. The cumulative histogram curve for the two datasets shows a slight underestimation of SMC in the ASAR product. This could be ascribed to the reduced resolution of ASAR WS and the use of VV polarization.

In this paper, a sensor-driven domain adaptation method is proposed for the classification of remote sensing images. The method aims at classifying an image where ground truth is not available exploiting the reference data acquired on a different but related image. This is done by taking advantage from a sensor-driven strategy that exploits the invariance of the measurements of some sensors on some classes for adaptation. This invariant property allows us to infer labels on a subset of unlabeled samples of the image that should be classified, thus introducing constrains on the adaptation process. The proposed method is based on two main steps: i) adaptation based on a sensor-driven label inference method for a subset of classes characterized by spatial invariant behaviour; and ii) adaptation based on machine learning for the remaining classes. The proposed method has been validated on 2 different datasets, where LiDAR data, hyperspectral images and high resolution optical images have been considered.

This paper presents a novel feature selection method for the analysis of hyperspectral images. The proposed method aims at selecting a subset of the original features that are both 1) relevant for the considered problem (i.e., preserve the functional relationship between input and output variables), and 2) invariant (stable) across different domains (i.e., minimize the data set shift among different domains). Domains can be associated with images collected on different areas or on the same area at different times. We propose a novel measure of domain stability, which evaluates the distance of the conditional distributions between the source and target domain. Such a measure is defined on the basis of kernel embeddings of conditional distributions and can be applied to both classification and regression problems. Experimental results show the effectiveness of the proposed method in selecting features with high generalization capabilities on the target domain.

This paper presents an adaptive framework for detection of changes of relevance occurring in image time series in a recursive way. With the availability of reference data for only one image pair from the time series (source domain), the proposed methodology employs change vector analysis in the 3-dimensional spherical domain to determine a decision region R associated with the change of relevance. Then, by exploiting the similarity among domains, the same kind of change can be detected by adapting R to the rest of image pairs belonging to the time series. The methodology was tested in a multispectral time series made up by TM-Landsat images marked by sequential deforestation activities in the Amazon with reference data. The quantitative analysis of the results indicates the soundness of the proposed approach.

This letter proposes a novel strategy for reducing the cost of in situ sample labeling for the definition of training sets by active learning (AL) in the framework of supervised classification of remote sensing images. AL methods define a training set according to an iterative procedure that at each iteration requires the labeling of a set of new samples selected by the classifier. The proposed strategy can be embedded in any AL method to identify the most informative area on the ground where focusing each AL iteration to reduce the overall cost (in terms of time) of labeling. To this end, at each iteration, the most uncertain unlabeled samples are initially identified. Then, the area on the ground (having a size predefined by the user) that has the highest spatial density of informative (i.e., uncertain and diverse) unlabeled samples is selected by the proposed strategy, and the AL technique is applied only to the samples of that area. This results in a decrease of the overall labeling cost with respect to that required by the use of a given technique in a standard way. Experimental results obtained by embedding the presented strategy in different literature active learning methods confirm its effectiveness.

The continuous melting of the ice at the Earth continental polar caps highlights the importance of an exhaustive study of the properties of the ice subsurface targets in order to provide a reliable analysis of their past and future evolution. Such study can be efficiently performed by automatically analyzing radargrams of the ice cross section acquired by radar sounder (RS) instruments. In this letter, we propose an automatic technique for a large-scale detection of the ice subsurface targets and the estimation of their properties (e.g., layered area thickness and bedrock scattering area) from radargrams acquired by RS operated at the Earth continental polar caps. This is done by using the parameters of the RS acquisition system combined with the output of an automatic image segmentation algorithm. The segmentation operation is applied to the radargrams after a preliminary processing phase aimed to emphasize the relevant subsurface targets. The segmentation criterion considers the radar signal backscattering properties and a model of the spatial distribution of the investigated targets that takes into account the effects of the wave propagation though the subsurface. Experimental results obtained on real radargrams acquired by an airborne RS in Antarctica confirm the effectiveness of the proposed technique.

This paper proposes a novel cost-sensitive active learning (CSAL) method to the definition of reliable training sets for the classification of remote sensing images with support vector machines. Unlike standard active learning (AL) methods, the proposed CSAL method redefines AL by assuming that the labeling cost of samples during ground survey is not identical, but depends on both the samples accessibility and the traveling time to the considered locations. The proposed CSAL method selects the most informative samples on the basis of three criteria: 1) uncertainty; 2) diversity; and 3) labeling cost. The labeling cost of the samples is modeled by a novel cost function that exploits ancillary data such as the road network map and the digital elevation model of the considered area. In the proposed method, the three criteria are applied in two consecutive steps. In the first step, the most uncertain samples are selected, whereas in the second step the uncertain samples that are diverse and have low labeling cost are chosen. In order to select the uncertain samples that optimize the diversity and cost criteria, we propose two different optimization algorithms. The first algorithm is defined on the basis of a sequential forward selection optimization strategy, whereas the second one relies on a genetic algorithm. Experimental results show the effectiveness of the proposed CSAL method compared to standard AL methods that neglect the labeling cost.

This paper presents a novel semisupervised kernel partial least squares (KPLS) algorithm for nonlinear feature extraction to tackle both land-cover classification and biophysical parameter retrieval problems. The proposed method finds projections of the original input data that align with the target variable (labels) and incorporates the wealth of unlabeled information to deal with low-sized or underrepresented data sets. The method relies on combining two kernel functions: the standard radial-basis-function kernel based on labeled information and a generative, i.e., probabilistic, kernel directly learned by clustering the data many times and at different scales across the data manifold. The construction of the kernel is very simple and intuitive: Two samples should belong to the same class if they consistently belong to the same clusters at different scales. The effectiveness of the proposed method is successfully illustrated in multi- and hyperspectral remote sensing image classification and biophysical parameter estimation problems. Accuracy improvements in the range between +5% and 15% over standard principal component analysis (PCA), +4% and 15% over kernel PCA, and +3% and 10% over KPLS are obtained on several images. The average gain in the root-mean-square error of +5% and reductions in bias estimates of +3% are obtained for biophysical parameter retrieval compared to standard PCA feature extraction.

In this paper, a novel iterative active learning technique based on self-organizing map (SOM) neural network and support vector machine (SVM) classifier is presented. The technique exploits the properties of the SVM classifier and of the SOM neural network to identify uncertain and diverse samples, to include in the training set. It selects uncertain samples from low-density regions of the feature space by exploiting the topological properties of the SOM. This results in a fast convergence also when the available initial training samples are poor. The effectiveness of the proposed method is assessed by comparing it with several methods existing in the literature using a toy data set and a color image as well as real multispectral and hyperspectral remote sensing images.

This paper aims at analyzing and comparing active learning (AL) and semisupervised learning (SSL) methods for the classification of remote sensing (RS) images. We present a literature review of the two learning paradigms and compare them theoretically and experimentally when addressing classification problems characterized by few training samples (w.r.t. the number of features) and affected by sample selection bias. Commonalities and differences are highlighted in the context of a conceptual framework used to describe the workflow of the two approaches. We point out advantages and disadvantages of the two approaches, delineating the boundary conditions on the applicability of the two paradigms with respect to both the amount and the quality of available training samples. Moreover, we investigate the integration of concepts that are in common between the two learning paradigms for improving state-of-the-art techniques and combining AL and SSL in order to jointly leverage the advantages of both approaches. In this framework, we propose a novel SSL algorithm that improves the progressive semisupervised support vector machine by integrating concepts that are usually considered in AL methods. We performed several experiments considering both synthetic and real multispectral and hyperspectral RS data, defining different classification problems starting from different initial training sets. The experiments are carried out considering classification methods based on support vector machines.

In this manuscript introduces a novel 4 dimensional tool for visualizing and analyzing long time series. The proposed tool is thought to be a support for an intuitive visual analysis of long time series and a starting point for the definition of novel methods for automatic image information mining in time series. The proposed tool is based on the mathematical smooth space curve known as helix. This choice aims at preserving the time neighborhood of acquisitions as a space neighborhood in the 4D representation (e.g., the proximity of subsequent days over the New Year). The impact and usefulness of the proposed tool is validated on a long time series of MODIS Terra Surface Reflectance Daily L2G Global (MOD09GA) products with a spatial resolution of 500m acquired from 1stJanuary 2005 to 31stDecember 2012 for a total of 2845images.

This paper presents the Radar for Icy Moons Exploration (RIME) instrument, which has been selected as payload for the JUpiter Icy moons Explorer (JUICE) mission. JUICE is the first Large-class mission chosen as part of the ESA's Cosmic Vision 2015-2025 programme, and is aimed to study Jupiter and to investigate the potentially habitable zones in the Galilean icy satellites. RIME is a radar sounder optimized for the penetration of Ganymede, Europa and Callisto up to a depth of 9 km in order to allow the study of the subsurface geology and geophysics of the icy moons and detect possible subsurface water. In this paper we present the main science goals of RIME, the main technical challenges for its development and for its operations, as well as the expected scientific returns.

This paper presents a novel active learning (AL) technique to drive relevance feedback in content based image retrieval (CBIR) from earth observation data archives. The proposed AL method aims at defining an effective set of relevant and irrelevant images with respect to the query image as small as possible. This is achieved on the basis of a joint evaluation of three criteria: i) uncertainty, ii) diversity and iii) density of images. The uncertainty and diversity criteria aims at choosing the most informative images in the archive, whereas the density criterion aims at selecting those that are representative of the underlying distribution of images in the archive. In the proposed AL method, the three criteria are applied in two consecutive steps. In the first step the most uncertain images are selected based on well-known margin sampling strategy. In the second step the images that are associated to high density regions in the archive and are diverse (i.e., distant) to each other are chosen from the most uncertain ones on the basis of a novel clustering based strategy. Experimental results show the effectiveness of the proposed AL method, particularly when a poor initial set of relevant and irrelevant images is available.

This paper presents a novel sequential cascade classification technique for automatically updating land-cover maps by classifying remote sensing image time series. We assume that a reliable training set is initially available only for one of the images (i.e., the source domain) in the time series, whereas it is not for an image being classified (i.e., the target domain). Unlike the standard cascade classification method, the proposed method aims at exploiting all the images in the time series acquired between the target and source domains to effectively classify the target domain. To this end, initially `pseudo' training sets of the images are defined by a multiple pairwise change detection based transfer learning strategy. Then, the target domain is classified by the proposed sequential cascade classification method, exploiting the temporal correlation between images. Experimental results obtained on a time series of Landsat multispectral images show the effectiveness of the proposed technique with respect to the standard cascade classification.

The sea level rise determined by the continuous increase in the global temperature calls for a quantitative investigation of the continental ice subsurface features and their dynamics. In the past decades, the study of these features has been carried out by manually analyzing radargrams acquired by airborne-mounted radar sounder (RS) instruments at the Earth polar caps. As RSs provide a very large amount of data, the main challenge to an exhaustive analysis of the ice subsurface is the efficient extraction of useful information contained in radargrams. To address this challenge, in this paper we propose an automatic classification system of the main ice subsurface features visible in radargrams, i.e., ice layered area, bedrock scattering area and noise regions. The system relies on the extraction of a set of discriminant features which are computed on the bases of a detailed analysis of the statistical properties of the radar signal and of the spatial distribution of the subsurface features. The features are then given as input to a machine learning classifier based on Support Vector Machine (SVM). The proposed system is validated on a dataset made up of several radargrams acquired by an airborne RS in Antarctica.

This paper addressed the change-detection problem in multitemporal hyperspectral remote sensing images (CD-HS). The concept of “change” in multitemporal hyperspectral images is analyzed from the viewpoint of single pixel spectral signal. A novel hierarchical change-detection approach is proposed by considering both the change magnitude and spectral change information, which aims to identify the change classes having discriminable spectral behaviors. The proposed method is developed in an unsupervised way thus to provide a solution for real CD-HS cases, for which reference samples are often not available. Experimental results obtained on multitemporal Hyperion hyperspectral images confirm the effectiveness of the proposed change-detection approach.

This paper presents a novel approach to multitemporal detection of primitives in very high resolution (VHR) SAR images. The proposed approach aims at exploiting the monotemporal as well as the multitemporal information in order to both: i) identify transitions in the state of primitives (i.e., detect changes); and ii) improve the monotemporal detection of primitives taking explicitly advantage of the temporal correlation. The performance of the multitemporal detector is evaluated on a time series of four TerraSAR-X images acquired over the city of Lüneburg in Germany. Experimental results confirm the effectiveness of the proposed approach.

This paper presents an approach to the detection of changed buildings using multitemporal Very High Resolution (VHR) Synthetic Aperture Radar (SAR) images. The proposed approach is based on two concepts i) the extraction of information on changes associated with increase and decrease of backscattering at the optimal building scale; and ii) the exploitation of the expected backscattering proprieties of buildings to detect new and fully destroyed buildings with their grade of reliability. Experimental results obtained on a dataset made up of two COSMO-SkyMed (CSK©) spotlight images acquired in 2009 over the city of L'Aquila (Italy) before and after an earthquake demonstrated that the proposed approach allows an effective identification of destroyed buildings.

Light detection and ranging (LiDAR) is one of the most efficient remote sensing technologies for the estimation of forest parameters. However, when acquired with a low laser sampling density, LiDAR data fail in providing accurate tree height measures. In order to address this issue, in this paper we propose a novel technique for the reconstruction of tree-top height based on the joint use of low-density LiDAR data and high resolution optical images. The proposed method is based on the following steps: i) detection of all the tree crowns present in the scene by fusing the two remotely sensed data sources; ii) reconstruction of the tree-top height for those crown hit by at least one LiDAR point; iii) estimation of the tree-top height for those crowns without LiDAR points. The proposed technique has been tested on a coniferous forest located in the Italian Alps. The experimental results points out the effectiveness of the proposed method.

A new approach to change detection in very high resolution remote sensing images based on morphological attribute profiles (APs) is presented. A multiresolution contextual transformation performed by APs allows the extraction of geometrical features related to the structures within the scene at different scales. The temporal changes are detected by comparing the geometrical features extracted from the image of each date. The experiments performed on panchromatic QuickBird images related to an urban area show the effectiveness of the proposed technique in detecting changes on the basis of the spatial morphology by preserving geometrical detail.

Extended attribute profiles, which are based on attribute filters, have recently been presented as efficient tools for spectral-spatial classification of remote sensing images. However, construction of these profiles usually requires manual selection of parameters for the corresponding attribute filters. In this letter, we present a technique to automatically build the extended attribute profiles with the standard deviation attribute based on the statistics of the samples belonging to the classes of interest. The methodology is tested on two widely used hyperspectral images and the results are found to be highly accurate.

Relevant component analysis has shown effective in metric learning. It finds a transformation matrix of the feature space using equivalence constraints. This paper explores this idea for constructing a feature metric (FM) and develops a novel semisupervised feature-selection technique for hyperspectral image classification. Two feature measures referred to as band correlation metric (BCM) and band separability metric (BSM) are derived for the FM. The BCM can measure the spectral correlation among the bands, while the BSM can assess the class discrimination capability of a single band. The proposed feature-metric-based affinity propagation (AP) (FM-AP) technique utilizes exemplar-based clustering, i.e., AP, to group bands from original spectral channels with the FM. Experimental results are conducted on two hyperspectral images and show the advantages of the proposed technique over traditional feature-selection methods.

This paper addresses change detection in multitemporal remote sensing images. After a review of the main techniques developed in remote sensing for the analysis of multitemporal data, the attention is focused on the challenging problem of change detection in very-high-resolution (VHR) multispectral images. In this context, we propose a framework that aims at defining a top-down approach to the design of the architecture of novel change-detection systems for multitemporal VHR images. The proposed framework explicitly models the presence of different radiometric changes on the basis of the properties of multitemporal images, extracts the semantic meaning of radiometric changes, identifies changes of interest with strategies designed on the basis of the specific application, and takes advantage of the intrinsic multiscale/multilevel properties of the objects and the high spatial correlation between pixels in a neighborhood. This framework defines guidelines for the development of a new generation of change-detection methods that can properly analyze multitemporal VHR images taking into account the intrinsic complexity associated with these data. In order to illustrate the use of the proposed framework, a real change-detection problem has been considered, which is described by a pair of VHR multispectral images acquired by the QuickBird satellite on the city of Trento, Italy. The proposed framework has been used for defining a system for change detection in the two images. Experimental results confirm the effectiveness of the developed system and the usefulness of the proposed framework.

The availability of very high resolution (VHR) synthetic aperture radar (SAR) images, which can be acquired by satellites over the same geographical area with short repetition interval, makes the development of effective unsupervised change detection (CD) techniques very important. This paper proposes a hierarchical approach to CD in VHR SAR images for addressing surveillance applications, where VHR data are acquired with high temporal resolution (e.g., one image every few days). The proposed approach is based on two concepts: exploitation of a multiscale technique for a preliminary detection of areas containing changes in backscattering at different scales (hot spots) and explicit modeling of the semantic meaning of changes by using both the intrinsic SAR image properties (e.g., acquisition geometry and scattering mechanisms) and the available prior information. In order to illustrate the effectiveness of the proposed approach, a problem of freight traffic surveillance is addressed considering two data sets. Each of them is made up of a pair of multitemporal VHR SAR images acquired by the COSMO-SkyMed (COnstellation of small Satellites for the Mediterranean basin Observation) constellation in spotlight mode. Each data set defines a complex CD problem due to both the presence of a variety of changes on the ground and the complexity of object backscattering. Experimental results point out the effectiveness of the proposed approach.

This paper proposes a novel change-detection-driven transfer learning (TL) approach to update land-cover maps by classifying remote-sensing images acquired on the same area at different times (i.e., image time series). The proposed approach requires that a reliable training set is available only for one of the images (i.e., the source domain) in the time series whereas it is not for another image to be classified (i.e., the target domain). Unlike other literature TL methods, no additional assumptions on either the similarity between class distributions or the presence of the same set of land-cover classes in the two domains are required. The proposed method aims at defining a reliable training set for the target domain, taking advantage of the already available knowledge on the source domain. This is done by applying an unsupervised-change-detection method to target and source domains and transferring class labels of detected unchanged training samples from the source to the target domain to initialize the target-domain training set. The training set is then optimized by a properly defined novel active learning (AL) procedure. At the early iterations of AL, priority in labeling is given to samples detected as being changed, whereas in the remaining ones, the most informative samples are selected from changed and unchanged unlabeled samples. Finally, the target image is classified. Experimental results show that transferring the class labels from the source domain to the target domain provides a reliable initial training set and that the priority rule for AL results in a fast convergence to the desired accuracy with respect to Standard AL.

Nowadays, the interest on the development of orbiting radar sounders for the observation of Earth polar areas is increasing. In this context, the analysis of the structure of the ice stratigraphy is of primary importance for the study of the past history and for the prediction of the evPolution of icy environments. However, as proven by planetary missions, orbiting radar sounders provide a huge amount of data. Thus, the development of automatic techniques for the analysis of these data is of fundamental importance for proper data exploitation. In this paper, we propose a novel method for the automatic detection of subsurface linear features from radar sounder data acquired in icy regions showing extended layering. The proposed method allows the estimation of the position of the linear features with subpixel accuracy. Moreover, each detected linear interface is treated as a single object which is completely described by the position of its points, the estimated local width, and the contrast. This allows the direct measurement of geometrical and radiometric parameters (e.g., slope angle and intensity) without the need of further postprocessing steps. This paper also proposes some measurements for deriving from the output of the proposed technique important variables that can characterize quantitatively the properties of the detected linear features (e.g., mean depth and mean intensity) and their distribution (e.g., number and density of layers). The effectiveness of the proposed method is confirmed by the results obtained on several radargrams acquired by the Shallow Radar on the North Pole of Mars.

The spaceborne synthetic aperture radar (SAR) systems Cosmo-SkyMed, TerraSAR-X, and TanDEM-X acquire imagery with very high spatial resolution (VHR), supporting various important application scenarios, such as damage assessment in urban areas after natural disasters. To ensure a reliable, consistent, and fast extraction of the information from the complex SAR scenes, automatic information extraction methods are essential. Focusing on the analysis of urban areas, which is of prime interest of VHR SAR, in this paper, we present a novel method for the automatic detection and 2-D reconstruction of building radar footprints from VHR SAR scenes. Unlike most of the literature methods, the proposed approach can be applied to single images. The method is based on the extraction of a set of low-level features from the images and on their composition to more structured primitives using a production system. Then, the concept of semantic meaning of the primitives is introduced and used for both the generation of building candidates and the radar footprint reconstruction. The semantic meaning represents the probability that a primitive belongs to a certain scattering class (e.g., double bounce, roof, facade) and has been defined in order to compensate for the lack of detectable features in single images. Indeed, it allows the selection of the most reliable primitives and footprint hypotheses on the basis of fuzzy membership grades. The efficiency of the proposed method is demonstrated by processing a 1-m resolution TerraSAR-X spotbeam scene containing flat- and gable-roof buildings at various settings. The results show that the method has a high overall detection rate and that radar footprints are well reconstructed, in particular for medium and large buildings.

One of the most critical problems that affect the analysis of orbiting radar sounder data is the presence of spurious surface clutter returns. These are due to off-nadir echoes related to surface topography which may be detected as (or mask) actual subsurface targets. The detection of such returns is usually carried out manually through a visual comparison between actual radargrams and surface clutter simulations obtained using available digital elevation models (DEMs). This is an inherently subjective and time-consuming task, which may reduce the scientific return of the data. In this paper, we address this problem by proposing a novel technique for the automatic detection of surface clutter returns in radar sounder data. The proposed method is made up of three steps: 1) the simulation of surface clutter returns using available DEMs; 2) the automatic coregistration between radargrams and simulations; and 3) the extraction of surface clutter returns from the coregistered radargrams. The coregistration step is performed in two phases: 1) a coarse registration based on the detection of the first return line on both input radargrams and 2) a fine registration based on B-spline deformation. The proposed technique is robust to radargram geometric deformations (e.g., due to ionospheric effects) and allows the generation of different types of outputs (e.g., coregistered simulations, binary clutter maps, and false-color compositions) that can both greatly support the scientific community in the manual analyses of radar sounder data and drive the development of reliable automatic methods for high level processing. The effectiveness of the proposed method is proven on two data sets acquired on different areas of Mars by the Shallow Radar instrument.

Morphological and attribute profiles have been proven to be effective tools to fuse spectral and spatial information for classification of remote sensing data. A wide range of filters (i.e., number of levels in the profiles) is usually necessary in order to properly model the spatial information in a remote sensing scene. A dense sampling of the values of the parameters of the filters generates profiles that have both a very large dimensionality (leading to the Hughes phenomenon in classification) and a high redundancy. In this paper, a novel iterative technique based on genetic algorithms (GAs) is proposed to automatically optimize the selection of the optimal features from the profiles. The selection of the filtered images that compose the profile is performed by dividing them into three classes corresponding to high, medium, and low importance. We propose to measure the importance (modeled in terms of discriminative power in the classification task) using a random forest classifier, which provides a rank for each feature with its model. Only the set of images associated with the highest importance is selected, i.e., preserved for classification. The proposed technique is applied to the features labeled with medium importance, whereas the images with the lowest importance are removed from the profile. This method is employed to classify three hyperspectral data sets achieving significantly high classification accuracy values. A parallel computing implementation has been developed in order to significantly reduce the time required for the run of the GAs.

Clustering is used for land-cover identification in remote sensing images when training data are not available. However, in many applications, it is often possible to collect a small number of labeled samples. To effectively exploit this small number of labeled samples combined with a multitude of the unlabeled data, we present a novel semisupervised clustering technique [incremental and decremental affinity propagation (ID-AP)] that incorporates labeled exemplars into the AP algorithm. Unlike standard semisupervised clustering methods, the proposed technique improves the performance by using both the labeled samples to adjust the similarity matrix and an ID-learning principle for unlabeled data selection and useless labeled samples rejection, respectively. This avoids both learning-bias and stability-plasticity dilemma. In order to assess the effectiveness of the proposed ID-AP technique, the experimental analysis was carried out on three different kinds of multispectral images including different percentages of labeled samples. In the analysis, we also studied the accuracy and the stability of two semisupervised clustering algorithms [i.e., constrainedk-means and semisupervised AP (SAP)] and one incremental semisupervised clustering algorithm (i.e., incremental SAP). Experimental results demonstrate that the proposed ID-AP technique adequately captures and takes full advantage of the intrinsic relationship between the labeled samples and unlabeled data, and produces better performance than the other considered methods.

Image classification usually requires the availability of reliable reference data collected for the considered image to train supervised classifiers. Unfortunately when time series of images are considered, this is seldom possible because of the costs associated with reference data collection. In most of the applications it is realistic to have reference data available for one or few images of a time series acquired on the area of interest. In this paper, we present a novel system for automatically classifying image time series that takes advantage of image(s) with an associated reference information (i.e., the source domain) to classify image(s) for which reference information is not available (i.e., the target domain). The proposed system exploits the already available knowledge on the source domain and, when possible, integrates it with a minimum amount of new labeled data for the target domain. In addition, it is able to handle possible significant differences between statistical distributions of the source and target domains. Here, the method is presented in the context of classification of remote sensing image time series, where ground reference data collection is a highly critical and demanding task. Experimental results show the effectiveness of the proposed technique. The method can work on multimodal (e.g., multispectral) images.

In this paper, a spatiocontextual unsupervised change detection technique for multitemporal, multispectral remote sensing images is proposed. The technique uses a Gibbs Markov random field (GMRF) to model the spatial regularity between the neighboring pixels of the multitemporal difference image. The difference image is generated by change vector analysis applied to images acquired on the same geographical area at different times. The change detection problem is solved using the maximum a posteriori probability (MAP) estimation principle. The MAP estimator of the GMRF used to model the difference image is exponential in nature, thus a modified Hopfield type neural network (HTNN) is exploited for estimating the MAP. In the considered Hopfield type network, a single neuron is assigned to each pixel of the difference image and is assumed to be connected only to its neighbors. Initial values of the neurons are set by histogram thresholding. An expectation-maximization algorithm is used to estimate the GMRF model parameters. Experiments are carried out on three-multispectral and multitemporal remote sensing images. Results of the proposed change detection scheme are compared with those of the manual-trial-and-error technique, automatic change detection scheme based on GMRF model and iterated conditional mode algorithm, a context sensitive change detection scheme based on HTNN, the GMRF model, and a graph-cut algorithm. A comparison points out that the proposed method provides more accurate change detection maps than other methods.

This paper proposes an approach to change detection in multitemporal very high geometrical resolution (VHR) SAR images for surveillance applications. The approach takes advantage of 3 concepts: i) multiscale representation for a preliminary detection of areas showing significant changes in backscattering between the two images (hot spots); ii) exploitation of prior information about typical usage of zones of interest in the area under control; and iii) definition of features and change detectors optimized for an effective detection of specific changes in each zone of interest. Here the proposed approach is designed for the solution of surveillance problems. A data set made up of a pair of multitemporal VHR SAR images acquired by the COSMO-SkyMed (CSK®) constellation in spotlight mode on the commercial port of Livorno (Italy) was used. Experimental results point out the effectiveness of the proposed approach.

This paper presents a novel system for automatically updating land-cover maps by classifying remote sensing image time series. The proposed system assumes that a reliable training set is available only for one of the images (i.e., the source domain) in the time series, whereas it is not for another image to be classified (i.e., the target domain). To effectively classify the target domain the proposed system includes two steps: i) low-cost definition of the training set for the target domain; and ii) target domain classification according to the Bayesian cascade decision rule that exploits the temporal correlation between domains. In the proposed system, the low cost training set for the target domain is defined on the basis of transfer and active learning methods, which also use the temporal dependence information between the domains. Experimental results obtained on a time series of Landsat multispectral images show the effectiveness of the proposed technique.

This paper presents a novel cost-sensitive active learning technique (CSAL) to define effective training sets for the classification of remote sensing images. Unlike the standard active learning methods, the proposed technique redefines AL by assuming that the labeling cost of samples when ground survey is used is not uniform and depends both on the samples accessibility and the traveling time to the considered locations. Accordingly, the proposed CSAL technique is based on the joint evaluation of three criteria for the selection of the most informative samples that have a low labeling cost: i) uncertainty, ii) diversity and iii) cost efficiency. The labeling cost of the samples is assessed by using ancillary data like the road map and the digital elevation model of the considered area. Experimental results show the effectiveness of the proposed CSAL method compared to the standard active learning methods that neglect the labeling cost.

This paper presents a novel unsupervised target-driven change detection procedure for analyzing multi-temporal remote sensing images, which is based on data transformation and similarity measures. The iteratively reweighted multivariate alteration detection (IR-MAD) technique is firstly used to separate the various change information into MAD components. Then, the similarity measures are used to automatically search for the target-related component according to a pre-defined target-driven rule. This procedure both takes advantage of the IR-MAD transformation in change detection and helps users to quickly locate the transformed component associated with their interesting change target. Experimental results obtained on multitemporal Landsat ETM+ data confirm the effectiveness of the proposed approach.

Orbiting radar sounders are very important instruments that can probe the subsurface of planetary bodies from orbit. In particular, they produce vertical profiles revealing the geological structure of the subsurface. As a consequence of their nadir-looking acquisition geometry, radargrams may be affected by spurious off-nadir echoes due to surface clutter. These returns represent a relevant problem, as they may be detected as (or mask) actual subsurface targets. Clutter is usually detected by means of manual visual comparison between radargrams and surface clutter simulations obtained using available digital elevation models (DEM). This procedure is inherently subjective and time-consuming, thus leading to a possible reduction of the scientific return of the data. This paper addresses this problem by proposing a novel technique for the automatic detection of surface clutter returns in radar sounder data which can both greatly support the scientific community in manual analysis of radar sounder data and drive the development of reliable automatic methods. The proposed method is made up of three steps: i) simulation of surface clutter returns using available digital elevation models, ii) automatic coregistration between radargrams and simulations, and iii) extraction of surface clutter returns from the coregistered radargrams. The technique has been tested on two large datasets acquired on different areas of Mars by the Shallow Radar (SHARAD) instrument. The results point out that the method is robust to radargram deformations (e.g., due to ionosphere effects) and that the coregistration process can achieve a very high accuracy.

When dealing with supervised target detection, the acquisition of labeled samples is one of the most critical phases: the samples must be yet representative of the class of interest, but must also be found among a vast majority of non-target examples. Moreover, the efficiency of the search is also an issue, since the samples labeled as background are not used by target detectors such as the support vector data description (SVDD). In this work we propose a competitive and effective approach to identify the most relevant training samples for one-class classification based on the use of an active learning strategy. The SVDD classifier is first trained with insufficient target examples. It is then used to detect the most informative samples to be labeled by a user through active learning techniques. By selecting unlabeled samples in a smart way and by adopting a diversity criterion, it is possible to obtain an accurate description of the class of interest with a relatively small number of training samples. The performance of the proposed method is illustrated in a change detection scenario and is validated by comparison with state-of-art active learning techniques originally developed for multiclass problems.

The rise in the temperature of the planet has a very negative impact on the subsurface dynamics of the Earth Polar Regions. This makes it crucial to investigate the features present in the ice subsurface. The analysis of these features is typically performed manually, by examining radargrams acquired by radar sounder (RS) instruments operated at the Polar Caps. However, in order to cope with the very large amount of data that RSs can acquire, it is necessary to develop data analysis techniques that can identify and extract subsurface features automatically. To address this problem, in this paper we propose a novel technique for the automatic estimation of the ice thickness and bedrock properties from RS data acquired in Antarctica. The proposed technique generates a statistical map of the subsurface by exploiting the statistical properties of the radar signal. Then, it applies a segmentation algorithm properly tuned to the characteristics of the investigated areas. In order to assess the effectiveness of the proposed technique, we analyzed its performance when applied to Multichannel Coherent Radar Depth Sounder (MCoRDS) data acquired in Antarctica.

Feature extraction is of paramount importance for an accurate classification of remote sensing images. Techniques based on data transformations are widely used in this context. However, linear feature extraction algorithms, such as the principal component analysis and partial least squares, can address this problem in a suboptimal way because the data relations are often nonlinear. Kernel methods may alleviate this problem only when the structure of the data manifold is properly captured. However, this is difficult to achieve when small-size training sets are available. In these cases, exploiting the information contained in unlabeled samples together with the available training data can significantly improve data description by defining an effective semisupervised nonlinear feature extraction strategy. We present a novel semisupervised Kernel Partial Least Squares (KPLS) algorithm for non-linear feature extraction. The method relies on combining two kernel functions: the standard RBF kernel using labeled information and a generative kernel directly learned by clustering the data. The effectiveness of the proposed method is successfully illustrated in multi- and hyper-spectral remote sensing image classification: accuracy improvements between +15 - 20% over standard PCA and +10% over advanced kernel PCA and KPLS for both images is obtained. Matlab code is available at http://isp.uv.es for the interested readers.

This paper presents a study on the retrieval of snowpack biophysical parameters in mountain areas from satellite remote sensing imagery. More in detail, the integration of new generation X-band Cosmo SkyMed SAR imagery and land surface temperature (LST) information derived from optical thermal remote sensing is investigated. First, a sensitivity analysis is carried out, in order to understand whether and to what extent the investigated remote sensing signals are sensitive to variations in different snowpack target parameters. Then, an advanced retrieval system based on the Support Vector Machine approach in a multilevel architecture is developed. Experiments carried out in a small valley in the eastern Alps during the winter 2010/11 point out the effectiveness of the combined use of X-band SAR and thermal satellite imagery for the characterization of snowpack parameters in terms of both accuracy on reference point measurements and capability to reproduce spatial patterns of the target variables.

This paper presents a novel batch mode active learning technique for solving remote sensing image classification problems. The proposed technique incorporates uncertainty, diversity and cluster assumption criteria to design the query function. The uncertainty criterion is implemented by taking into account the properties of the support vector machine classifiers. The diversity and cluster assumption criteria are defined by exploiting the properties of the self-organizing map neural networks. To assess the effectiveness of the proposed method, we compared it with several other active learning methods existing in the remote sensing literature by using both multispectral and hyperspectral remote sensing data sets. Experimental results confirmed the effectiveness of the proposed technique.

Dealing with a high number of features belonging to different types of data such as Hyperspectral image and Morphological Attribute Profiles (MAPs) might lead to a poor predictive performance of the classifier and hence low final accuracies of classification. This is due to the Hughes effect that consistently decreases the power of prediction of the classifier, in case of a limited and fixed number of training samples. In order to reduce the number of features and only keeping those which are more informative, a novel supervised feature selection technique based on GAs and the measure of the relevance of the features is presented in this work. Moreover, the effectiveness of the proposed technique was demonstrated by experimenting on an optical remote sensed dataset.

In the framework of the monitoring of structures and infrastructures from environmental disasters, the COSMO-SkyMed constellation has a huge potential, thanks to up to metric spatial resolution, short revisit time, and the day/night all-weather acquisition capability ensured by SAR. This paper focuses on the scientific results of the project “Development and validation of multitemporal image analysis methodologies for multirisk monitoring of critical structures and infrastructures,” funded by the Italian Space Agency. Several change-detection, data-fusion, and feature-extraction techniques, which were developed and experimentally validated in the project for COSMO-SkyMed imagery and for their integration with other data sources (including very high resolution optical data), are described and examples of processing results are discussed.

In this letter, we present a novel batch-mode active learning technique for solving multiclass classification problems by using the support vector machine classifier with the one-against-all architecture. The uncertainty of each unlabeled sample is measured by defining a criterion which not only considers the smallest distance to the decision hyperplanes but also takes into account the distances to other hyperplanes if the sample is within the margin of their decision boundaries. To select batch of most uncertain samples from all over the decision region, the uncertain regions of the classifiers are partitioned into multiple parts depending on the number of geometrical margins of binary classifiers passing on them. Then, a balanced number of most uncertain samples are selected from each part. To minimize the redundancy and keep the diversity among these samples, the kernelk-means clustering algorithm is applied to the set of uncertain samples, and the representative sample (medoid) from each cluster is selected for labeling. The effectiveness of the proposed method is evaluated by comparing it with other batch-mode active learning techniques existing in the literature. Experimental results on two different remote sensing data sets confirmed the effectiveness of the proposed technique.

This paper addresses the problem of land-cover map updating by classification of multitemporal remote-sensing images in the context of domain adaptation (DA). The basic assumptions behind the proposed approach are twofold. The first one is that training data (ground reference information) are available for one of the considered multitemporal acquisitions (source domain) whereas they are not for the other (target domain). The second one is that multitemporal acquisitions (i.e., target and source domains) may be characterized by different sets of classes. Unlike other approaches available in the literature, the proposed DA Bayesian classifier based on maximum a posteriori decision rule (DA-MAP) automatically identifies whether there exist differences between the set of classes in the target and source domains and properly handles these differences in the updating process. The proposed method was tested in different scenarios of increasing complexity related to multitemporal image classification. Experimental results on medium-resolution and very high resolution multitemporal remote-sensing data sets confirm the effectiveness and the reliability of the proposed DA-MAP classifier.

The detection of multiple changes (i.e., different kinds of change) in multitemporal remote sensing images is a complex problem. When multispectral images havingBspectral bands are considered, an effective solution to this problem is to exploit all available spectral channels in the framework of supervised or partially supervised approaches. However, in many real applications, it is difficult/impossible to collect ground truth information for either multitemporal or single-date images. On the opposite, unsupervised methods available in the literature are not effective in handling the full information present in multispectral and multitemporal images. They usually consider a simplified subspace of the original feature space having small dimensionality and, thus, characterized by a possible loss of change information. In this paper, we present a framework for the detection of multiple changes in bitemporal and multispectral remote sensing images that allows one to overcome the limits of standard unsupervised methods. The framework is based on the following: 1) a compressed yet efficient 2-D representation of the change information and 2) a two-step automatic decision strategy. The effectiveness of the proposed approach has been tested on two bitemporal and multispectral data sets having different properties. Results obtained on both data sets confirm the effectiveness of the proposed approach.

This paper presents a novel iterative active learning (AL) technique aimed at defining effective multitemporal training sets to be used for the supervised detection of land-cover transitions in a pair of remote sensing images acquired on the same area at different times. The proposed AL technique is developed in the framework of the Bayes' rule for compound classification. At each iteration, it selects the pair of spatially aligned unlabeled pixels in the two images that are classified with the maximum uncertainty. These pixels are then labeled by an external supervisor and included in the training set. The uncertainty of a pair of pixels is assessed by the joint entropy defined by considering two possible different simplifying assumptions: 1) class-conditional independence and 2) temporal independence between multitemporal images. Accordingly, different algorithms are introduced. The proposed joint-entropy-based AL algorithms for compound classification are compared with each other and with a marginal-entropy-based AL technique (in which the entropy is computed separately on single-date images) applied to the postclassification comparison method. The experimental results obtained on two multispectral and multitemporal data sets show the effectiveness of the proposed technique.

Radar sounders operating on satellite platforms (e.g., radar sounding missions at Mars) provide a huge amount of data that currently are mostly analyzed by means of manual investigations. This calls for the development of novel techniques for the automatic extraction of information from sounder signals that could greatly support the scientific community. Such a topic has not been addressed sufficiently in the literature. This paper provides a contribution to fill this gap by presenting both 1) a study of the theoretical statistical properties of radar sounder signals, and 2) two novel techniques for the automatic analysis of sounder radargrams. The main goal of the study is the identification of statistical distributions that can accurately model the amplitude fluctuations of different subsurface targets. This is fundamental for the understanding of signal properties and for the definition of automatic data analysis techniques. The results of such a study drive the development of two novel techniques for 1) the generation of subsurface feature maps, and 2) the automatic detection of the deepest scattering areas visible in the radargrams. The former produces for each radargram a map showing which areas have high probability to contain relevant subsurface features. The latter exploits a region-growing approach properly defined for the analysis of radargrams to identify and compose the basal scattering areas. Experimental results obtained on Shallow Radar data acquired on Mars confirm the effectiveness of the proposed techniques.

This paper presents a novel technique for addressing domain adaptation (DA) problems with active learning (AL) in the classification of remote sensing images. DA models the important problem of adapting a supervised classifier trained on a given image (source domain) to the classification of another similar but not identical image (target domain) acquired on a different area. The main idea of the proposed approach is iteratively labeling and adding to the training set the minimum number of the most informative samples from the target domain, while removing the source-domain samples that do not fit with the distributions of the classes in the target domain. In this way, the classification system exploits already available information, i.e., the labeled samples of source domain, in order to minimize the number of target domain samples to be labeled, thus reducing the cost associated to the definition of the training set for the classification of the target domain. In addition, we define a convergence criterion that allows the technique to stop the iterative AL process on the target domain without relying on the availability of a test set for it. This is an important contribution, as in operational applications, it is not realistic to assume that a test set for the target domain is available. Experimental results obtained in the classification of very high resolution and hyperspectral images confirm the effectiveness of the proposed technique.

In this paper an unsupervised approach to multiple change detection based on adaptive thresholding is proposed. The method is developed in the polar framework for change vector analysis in multispectral images recently presented in the literature. According to the properties of spectral change vectors in the polar framework a procedure is presented based on the following steps: i) identification of unchanged and changed pixels along the magnitude variable (different kinds of change are treated as being a single class); ii) isolation of different kinds of changes along the direction variable; iii) refinement of the decision threshold along the magnitude variable according to the properties of each kind of detected change considered independently. The proposed method is validated on a pair of multitemporal images acquired by the Landsat-5 satellite including 3 kinds of changes.

This paper addresses the problem of change detection (CD) in very high geometrical resolution (VHR) multitemporal images. In this context, we propose a conceptual framework that aims at defining: i) a taxonomy of different radiometric changes occurring when dealing with VHR remote sensing images; and ii) a global approach to the definition of the architecture of effective CD methods for VHR images. This framework defines precise guidelines for the development of a new generation of CD methods that can properly analyze multitemporal VHR images. The proposed framework is illustrated in the definition of a change detection method for the solution of a specific real problem related to the analysis of multitemporal QuickBird images.

This paper presents a novel active learning (AL) technique for the compound classification of multitemporal remote-sensing images for the detection of land-cover transitions. The proposed AL technique is based on the selection of unlabeled pairs of samples that have maximum uncertainty on their labels assigned by a classifier implemented according to the Bayes rule for compound classification. Uncertainty of a pair of samples is assessed by joint entropy defined on the basis of two different simplifying assumptions: i) class-conditional independence, and ii) temporal independence between multitemporal images. Accordingly, two algorithms for the proposed joint entropy based AL technique are introduced. The proposed joint entropy based AL algorithms are compared to each other and with a marginal entropy (entropy computed separately on single-date images) based AL technique. Experimental results obtained on two multispectral images show the effectiveness of the proposed technique.

Nowadays, planetary radar sounding missions operating at Mars are providing a huge amount of data which are mostly analyzed by means of manual investigations. The development of techniques for the automatic extraction of information from radar sounder data is thus a crucial issue for a proper data exploitation from the scientific community. As an important step to the definition of automatic methods that can be implemented in a more general framework for the analysis of radar sounder data, in this paper we present a study of the statistical properties of radar sounder signals. The statistical distributions of the amplitude of the samples related to different subsurface targets are analyzed in order to define a suitable fitting model. Exploiting the results of such a study, we also propose a novel automatic technique capable to extract the deepest scattering area visible in the radargrams. The basal scattering area is composed using an advanced region-growing approach properly defined for the analysis of radargrams. The obtained regions are kept or discarded according to the statistical distribution of their samples. Experimental results obtained on data acquired by the Shallow Radar (SHARAD) on Mars confirm the effectiveness of the proposed approach.

In this paper a general approach based on morphological connected filters for the spatial simplification of very high resolution remote sensing images is introduced. In greater detail, the proposed approach is made up of two steps: i) the selection of the parameters defining the connected filters driven by the information available on the scene and on the specific application; and ii) the application of the tuned filter to the input image. This work aims at: i) explicitly delineating the characteristic of an approach for the spatial simplification of images based on connected filters; ii) defining a general architecture suitable for the analysis in different scenarios modeling common different operative conditions; iii) giving guidelines for the automation of the simplification process according to different operational settings; iv) qualitatively evaluating the application of the proposed approach on a real data set in different scenarios.

This paper presents a novel hybrid approach to the estimation of biophysical parameters from remotely sensed data. This approach integrates theoretical analytical models and empirical models based on field reference samples to increase the reliability and the accuracy of the estimation. The estimation process is modeled by two terms: the first one expresses the relationship between the input features and the target biophysical variable according a theoretical model based on the physics of the considered problem; the second one corrects the deviation between theoretical model estimates and true target values according to an empirical data-driven model. The latter is derived by exploiting the available (typically few) field reference samples. In this way the robustness and generality of theoretical model based estimates, which stem from the rigorous theoretical foundation, is preserved, while the bias and imprecision (due to simplifications in the analytical formulations of the model with respect to the real estimation process) are reduced. Results achieved for the specific application of soil moisture estimation from microwave remotely sensed data with two different correction strategies are reported. These results show the effectiveness and the potentiality of the proposed integration approach.

In this work, fully polarimetric RADARSAT2 SAR images and advanced feature extraction strategies are investigated for improving the retrieval of soil moisture content in Alpine meadows and pastures. More in detail, standard Intensity & Phase polarimetric features, polarimetric decompositions and general purpose feature extraction strategies are exploited in combination with a sequential forward selection to increase the accuracy of the system. The capability of the system to provide spatially and temporally distributed estimates of soil moisture is also addressed by using different satellite acquisitions. The achieved results indicate that the polarimetric information in the SAR data, if properly exploited, allows one to improve the estimation of soil moisture content in the investigated mountain area. Concerning the mapping of the target variable, the analysis of the results suggest that the proposed estimation system is promising and effectively maps the soil moisture status both in time and space.

We present a novel technique for addressing domain adaptation problems in the classification of remote sensing images with active learning. Domain adaptation is the important problem of adapting a supervised classifier trained on a given image (source domain) to the classification of another similar (but not identical) image (target domain) acquired on a different area, or on the same area at a different time. The main idea of the proposed approach is to iteratively labeling and adding to the training set the minimum number of the most informative samples from target domain, while removing the source-domain samples that does not fit with the distributions of the classes in the target domain. In this way, the classification system exploits already available information, i.e., the labeled samples of source domain, in order to minimize the number of target domain samples to be labeled, thus reducing the cost associated to the definition of the training set for the classification of the target domain. Experimental results obtained in the classification of a hyperspectral image confirm the effectiveness of the proposed technique.

With the recent advent of very high resolution (VHR) spaceborne synthetic aperture radar (SAR) sensors such as TerraSAR-X and COSMO-SkyMed, the potential to use SAR simulators is increasing. In this letter, we propose a novel radar imaging simulator that is relatively simple to implement and that finds a balance between accuracy and efficiency. The main goal of the proposed method is to obtain a precise simulation of the geometry of objects in SAR images rather than a detailed radiometric simulation. The simulator is based on an extended ray-tracing procedure to determine which surfaces of a generic object contribute to the backscatter. The backscatter contributions are calculated by means of a Lambertian-specular mixture model. The simulator has already been employed successfully in a methodology for 3-D reconstruction of man-made objects from single detected VHR SAR imagery. Here, we illustrate its work on two rather different structures, a rectangular gable-roof building and an Egyptian pyramid.

In this letter, we study empirically the relation between the double-bounce effect of buildings in very high resolution (VHR) synthetic aperture radar (SAR) and the orientation angle for two different ground materials (i.e., asphalt and grass) by analyzing two different TerraSAR-X VHR spaceborne SAR images. Furthermore, we compare our empirical results with the simulations obtained using theoretical electromagnetic models. In order to deal with slightly rough surfaces, we also present a novel model for double-bounce scattering based on the small-perturbation method. We show that the double-bounce effect results in different power signatures, depending on the type of the building and the surrounding ground properties. Finally, we discuss the reliability of theoretical models for predicting the double-bounce power for the analyzed data sets. The models can predict the general behavior of the double bounce but lack in calculating the accurate double-bounce radar cross section reliably.

In this letter, a technique based on independent component analysis (ICA) and extended morphological attribute profiles (EAPs) is presented for the classification of hyperspectral images. The ICA maps the data into a subspace in which the components are as independent as possible. APs, which are extracted by using several attributes, are applied to each image associated with an extracted independent component, leading to a set of extended EAPs. Two approaches are presented for including the computed profiles in the analysis. The features extracted by the morphological processing are then classified with an SVM. The experiments carried out on two hyperspectral images proved the effectiveness of the proposed technique.

This letter presents an experimental analysis of the application of the ε-insensitive support vector regression (SVR) technique to soil moisture content estimation from remotely sensed data at field/basin scale. SVR has attractive properties, such as ease of use, good intrinsic generalization capability, and robustness to noise in the training data, which make it a valid candidate as an alternative to more traditional neural-network-based techniques usually adopted in soil moisture content estimation. Its effectiveness in this application is assessed by using field measurements and considering various combinations of the input features (i.e., different active and/or passive microwave measurements acquired using various sensor frequencies, polarizations, and acquisition geometries). The performance of the SVR method (in terms of estimation accuracy, generalization capability, computational complexity, and ease of use) is compared with that achieved using a multilayer perceptron neural network, which is considered as a benchmark in the addressed application. This analysis provides useful indications for building soil moisture estimation processors for upcoming satellites or near-real-time applications.

This paper provides an overview of the Europa Jupiter System Mission (EJSM) and of its scientific objectives, focusing the attention on the subsurface radar (SSR) instrument included in the model payload of the Jupiter Ganymede Orbiter (JGO). The SSR instrument is a radar sounder system at low frequency (HF/VHF band) designed to penetrate the surface of Ganymede icy moon of Jupiter for performing a subsurface analysis with a relatively high range resolution. This active instrument is aimed at acquiring information on the Ganymede (and partially on the Callisto during flybys) shallow subsurface. The paper addresses the main issues related to SSR, presenting its scientific goals, describing the concept and the design procedure of the instrument, and illustrating the signal processing techniques. Despite the fact that SSR can be defined on the basis of the heritage of the Mars Advanced Radar for Subsurface and Ionosphere Sounding (MARSIS) and SHAllow RADar (SHARAD) instruments currently operating at Mars, the EJSM mission poses additional scientific and technical challenges for its design: 1) the presence of a relevant Jupiter radio emission (which is very critical because it has a significant power spectral density in proximity of the expected SSR central frequency); 2) the properties of the subsurface targets, which are different from those of the Mars subsurface; 3) the different orbit conditions; and 4) the limited available resources (in terms of mass, power, and downlink data rate). These challenges are analyzed and discussed in relation to the design of the instrument in terms of: 1) choice of the central frequency and the bandwidth; 2) signal-to-noise ratio (SNR); 3) signal-to-clutter ratio (SCR); and 4) definition of the synthetic aperture processing. Finally, the procedure defined for SSR performance assessment is described and illustrated with some numerical examples.

Forest inventories are important tools for the management of forests. In this context, the estimation of the tree stem volume is a key issue. In this paper, we present a system for the estimation of forest stem diameter and volume at individual tree level from multireturn light detection and ranging (LIDAR) data. The proposed system is made up of a preprocessing module, a LIDAR segmentation algorithm (aimed at retrieving tree crowns), a variable extraction and selection procedure, and an estimation module based on support vector regression (SVR) (which is compared with a multiple linear regression technique). The variables derived from LIDAR data are computed from both the intensity and elevation channels of all available returns. Three different methods of variable selection are analyzed, and the sets of variables selected are used in the estimation phase. The stem volume is estimated with two methods: 1) direct estimation from the LIDAR variables and 2) combination of diameters and heights estimated from LIDAR variables with the species information derived from a classification map according to standard height/diameter relationships. Experimental results show that the system proposed is effective and provides high accuracies in both the stem volume and diameter estimations. Moreover, this paper provides useful indications on the effectiveness of SVR with LIDAR in forestry problems.

This paper investigates different batch-mode active-learning (AL) techniques for the classification of remote sensing (RS) images with support vector machines. This is done by generalizing to multiclass problem techniques defined for binary classifiers. The investigated techniques exploit different query functions, which are based on the evaluation of two criteria: uncertainty and diversity. The uncertainty criterion is associated to the confidence of the supervised algorithm in correctly classifying the considered sample, while the diversity criterion aims at selecting a set of unlabeled samples that are as more diverse (distant one another) as possible, thus reducing the redundancy among the selected samples. The combination of the two criteria results in the selection of the potentially most informative set of samples at each iteration of the AL process. Moreover, we propose a novel query function that is based on a kernel-clustering technique for assessing the diversity of samples and a new strategy for selecting the most informative representative sample from each cluster. The investigated and proposed techniques are theoretically and experimentally compared with state-of-the-art methods adopted for RS applications. This is accomplished by considering very high resolution multispectral and hyperspectral images. By this comparison, we observed that the proposed method resulted in better accuracy with respect to other investigated and state-of-the art methods on both the considered data sets. Furthermore, we derived some guidelines on the design of AL systems for the classification of different types of RS images.

In this paper, we propose a simple, fast, and reliable active-learning technique for solving remote sensing image classification problems with support vector machine (SVM) classifiers. The main property of the proposed technique consists in its robustness to biased (poor) initial training sets. The presented method considers the 1-D output space of the classifier to identify the most uncertain samples whose labeling and inclusion in the training set involve a high probability to improve the classification results. A simple histogram-thresholding algorithm is used to find out the low-density (i.e., under the cluster assumption, the most uncertain) region in the 1-D SVM output space. To assess the effectiveness of the proposed method, we compared it with other active-learning techniques proposed in the remote sensing literature using multispectral and hyperspectral data. Experimental results confirmed that the proposed technique provided the best tradeoff among robustness to biased (poor) initial training samples, computational complexity, classification accuracy, and the number of new labeled samples necessary to reach convergence.

In this paper a framework for the detection of multiple changes in multitemporal and multispectral remote sensing images is presented. The framework is based on: i) a compressed yet efficient (i.e., nearly lossless) 2-dimensional (2D) representation of the change information; and ii) a 2-step automatic decision strategy. At first, the original BD feature space to be explored for the solution of the change-detection (CD) problem is compressed to a 2D space in which the change information is clearly represented; then, the retrieved 2D space is explored for extracting in an automatic way the different kinds of change, thus generating the CD map. This procedure is conducted by applying a 2-step decision strategy based on the Bayes decision theory. Results obtained on a Landsat-5 and a QuickBird data sets confirm the effectiveness of the proposed approach in both representing the information in the 2D space and generating the CD map.

Information on the impact of catastrophic events (e.g. earthquakes) can be derived from suitable satellite imagery by comparing data from a chosen reference before the event (pre-event) to imagery acquired shortly after the event (post-event). In this paper, we propose a novel method that detects buildings destroyed in an earthquake using pre-event very high resolution (VHR) multispectral and post-event detected VHR synthetic aperture radar (SAR) imagery. The core concept of the proposed method is the evaluation of the presence of the predicted undamaged building SAR signature in the post-event SAR scene. The decision if a building belongs to the damaged or undamaged building class is performed with a Bayesian classifier, trained either in a supervised or unsupervised manner. We show the results of the proposed method using VHR TerraSAR-X and COSMO-SkyMed, as well as VHR optical data for a subset of the town of Yingxiu, China, which was heavily damaged in the 2008 Sichuan earthquake.

This paper addresses the problem of change detection (CD) in very high geometrical resolution (VHR) multitemporal images. In this context, we propose a general conceptual framework that aims at giving: i) a taxonomy of different radiometric changes occurring when dealing with VHR remote sensing images; and ii) a global approach to the definition of the architecture of effective CD methods for VHR images. This framework represents a first step for defining effective guidelines for the development of a new generation of CD methods that can be properly analyze multitemporal VHR images taking into account the intrinsic complexity associated with these data.

This paper addresses the recent trends in machine learning methods for the automatic classification of remote sensing (RS) images. In particular, we focus on two new paradigms: semisupervised and active learning. These two paradigms allow one to address classification problems in the critical conditions where the available labeled training samples are limited. These operational conditions are very usual in RS problems, due to the high cost and time associated with the collection of labeled samples. Semisupervised and active learning techniques allow one to enrich the initial training set information and to improve classification accuracy by exploiting unlabeled samples or requiring additional labeling phases from the user, respectively. The two aforementioned strategies are theoretically and experimentally analyzed considering SVM-based techniques in order to highlight advantages and disadvantages of both strategies.

The development of methods for the automatic detection and reconstruction of building radar footprints fromsingle very high resolution (VHR) synthetic aperture radar (SAR) images is a difficult task for two main reasons: i) the very high complexity of VHR SAR images; and ii) the need to develop efficient algorithms that can be applied to large images in order to be used in real applications. In this paper we present a novel method for automatic building detection, which also reconstructs the 2D radar footprint of the detected buildings. The method is based on the extraction of a set of low-level features from the images and on their combination in more structured primitives. Then the semantic meaning of primitives is used for the definition of building candidates and for the radar footprint reconstruction. In order to process large VHR SAR images, the method has been implemented on a computer cluster. We demonstrate the effectiveness of the method using a large TerraSAR-X spotlight scene.

In this paper a novel method for registration of multitemporal very high geometrical resolution (VHR) remote sensing images is presented. It relies on the extraction of a large set of control points (CPs) used for the estimation of a disparity map exploited for the registration process. CPs are automatically identified in both the images through the estimation and analysis of the distribution of registration noise (RN) and used together with an interpolation procedure in the definition of the disparity map. This map contains for each pixel the estimated value of the displacement between the reference and the moving image. The warping of the moving image is performed according to the disparity map by using thin plate spline interpolation. Results obtained on simulated and real VHR data confirm the validity of the proposed technique, which is effective both in identifying CPs and in performing the image alignment.

In this paper we investigate the combined use of morphological attribute filters and feature extraction techniques for the classification of a high resolution hyperspectral image. In greater detail, we propose to model the spatial information with Extended Attribute Profiles computed on the hyperspectral data and to reduce the high dimensionality of the morphological features computed (which show a high degree of redundancy) with feature extraction techniques. The features extracted are analyzed by two classifiers. The experimental analysis was carried out on a high resolution hyperspectral image acquired by the airborne sensor ROSIS-03 on the University of Pavia, Italy. The obtained results compared to those obtained without feature reduction proved the importance of the application of a stage of feature extraction in the process.

This paper proposes to model the critical issue of the choice of the free parameters of a supervised non-linear regression technique (the so called model selection issue) as a multiobjective optimization problem. In this framework, the multi-objective function is made up of a set of two or more quality metrics (e.g., MSE, R2, etc.) computed on the test (or validation) samples. A set of solutions is derived according to the concept of Pareto optimality. The advantages of the proposed approach with respect to the traditional ones (which typically optimize a single scalar metric) are mainly two: (1) the capability to derive solutions which jointly optimize the set of metrics considered and represent different possible optimal tradeoffs among them; and (2) the possibility for the user to effectively select the model that optimizes the requirements of the specific retrieval problem. Results achieved for the specific application of soil moisture estimation from microwave remotely sensed data with the Support Vector Regression (SVR) technique are reported. These results show the effectiveness of the proposed approach.

In this letter, we investigate the effects of pansharpening (PS) applied to multispectral (MS) multitemporal images in change-detection (CD) applications. Although CD maps computed from pansharpened data show an enhanced spatial resolution, they can suffer from errors due to artifacts induced by the fusion process. The rationale of our analysis consists in understanding to which extent such artifacts can affect spatially enhanced CD maps. To this end, a quantitative analysis is performed which is based on a novel strategy that exploits similarity measures to rank PS methods according to their impact on CD performance. Many multiresolution fusion algorithms are considered, and CD results obtained from original MS and from spatially enhanced data are compared.

This paper addresses pattern classification in the framework of domain adaptation by considering methods that solve problems in which training data are assumed to be available only for a source domain different (even if related) from the target domain of (unlabeled) test data. Two main novel contributions are proposed: 1) a domain adaptation support vector machine (DASVM) technique which extends the formulation of support vector machines (SVMs) to the domain adaptation framework and 2) a circular indirect accuracy assessment strategy for validating the learning of domain adaptation classifiers when no true labels for the target--domain instances are available. Experimental results, obtained on a series of two-dimensional toy problems and on two real data sets related to brain computer interface and remote sensing applications, confirmed the effectiveness and the reliability of both the DASVM technique and the proposed circular validation strategy.

Rapid damage assessment after natural disasters (e.g., earthquakes) and violent conflicts (e.g., war-related destruction) is crucial for initiating effective emergency response actions. Remote-sensing satellites equipped with very high spatial resolution (VHR) multispectral and synthetic aperture radar (SAR) imaging sensors can provide vital information due to their ability to map the affected areas with high geometric precision and in an uncensored manner. In this paper, we present a novel method that detects buildings destroyed in an earthquake using pre-event VHR optical and post-event detected VHR SAR imagery. The method operates at the level of individual buildings and assumes that they have a rectangular footprint and are isolated. First, the 3-D parameters of a building are estimated from the pre-event optical imagery. Second, the building information and the acquisition parameters of the VHR SAR scene are used to predict the expected signature of the building in the post-event SAR scene assuming that it is not affected by the event. Third, the similarity between the predicted image and the actual SAR image is analyzed. If the similarity is high, the building is likely to be still intact, whereas a low similarity indicates that the building is destroyed. A similarity threshold is used to classify the buildings. We demonstrate the feasibility and the effectiveness of the method for a subset of the town of Yingxiu, China, which was heavily damaged in the Sichuan earthquake of May 12, 2008. For the experiment, we use QuickBird and WorldView-1 optical imagery, and TerraSAR-X and COSMO-SkyMed SAR data.

Experimental airborne synthetic aperture radar (SAR) systems achieve spatial resolutions of approximately 10 cm, whereas the new spaceborne very high spatial resolution (VHR) SAR sensors onboard the TerraSAR-X and COSMO-SkyMed satellites achieve spatial resolutions down to 1 m. In VHR SAR data, features from individual urban structures (i.e., buildings) can be identified by their characteristic settings in urban settlement patterns. In this paper, we present a novel concept for the height estimation of generic man-made structures from single detected SAR data. The proposed approach is based on the definition of a hypothesis on the height of the building and on the simulation of a SAR image for testing that hypothesis. A matching procedure is applied between the estimated and the actual SAR image in order to test the height hypothesis. The process is iterated for different height assumptions until the matching function is optimized, and thus, the building height is estimated. The efficiency of the proposed method is demonstrated on a set of 40 flat- and gable-roof buildings using two submeter VHR airborne and two 1-m resolution TerraSAR-X SAR scenes all acquired from the same residential area in Dorsten, Germany. The results show that, in the absence of string disturbing effects, the method is able to estimate the height of flat- and gable-roof buildings in the submeter data to the order of a meter, while the accuracy for the meter resolution spaceborne data is lower but still sufficient to estimate the number of floors of a building.

Remote sensing image classification constitutes a challenging problem since very few labeled pixels are typically available from the analyzed scene. In such situations, labeled data extracted from other images modeling similar problems might be used to improve the classification accuracy. However, when training and test samples follow even slightly different distributions, classification is very difficult. This problem is known assample selection bias. In this paper, we propose a new method to combine labeled and unlabeled pixels to increase classification reliability and accuracy. A semisupervised support vector machine classifier based on the combination of clustering and themean mapkernel is proposed. The method reinforces samples in the same cluster belonging to the same class by combining sample and cluster similarities implicitly in the kernel space. Asoftversion of the method is also proposed where only the most reliable training samples, in terms of likelihood of the image data distribution, are used. Capabilities of the proposed method are illustrated in a cloud screening application using data from the MEdium Resolution Imaging Spectrometer (MERIS) instrument onboard the European Space Agency ENVISAT satellite. Cloud screening constitutes a clear example of sample selection bias since cloud features change to a great extent depending on the cloud type, thickness, transparency, height, and background. Good results are obtained and show that the method is particularly well suited for situations where the available labeled information does not adequately describe the classes in the test data.

This paper presents two semisupervised one-class support vector machine (OC-SVM) classifiers for remote sensing applications. Inone-classimage classification, one tries to detect pixels belonging to one of the classes in the image and reject the others. When few labeled pixels of only one class are available, obtaining a reliable classifier is a difficult task. In the particular case of SVM-based classifiers, this task is even harder because the free parameters of the model need to be finely adjusted, but no clear criterion can be adopted. In order to improve the OC-SVM classifier accuracy and alleviate the problem of free-parameter selection, the information provided by unlabeled samples present in the scene can be used. In this paper, we present two state-of-the-art algorithms for semisupervised one-class classification for remote sensing classification problems. The first proposed algorithm is based on modifying the OC-SVM kernel by modeling the data marginal distribution with the graph Laplacian built with both labeled and unlabeled samples. The second one is based on a simple modification of the standard SVM cost function which penalizes more the errors made when classifying samples of the target class. The good performance of the proposed methods is illustrated in four challenging remote sensing image classification scenarios where the goal is to detect one of the classes present on the scene. In particular, we present results for multisource urban monitoring, hyperspectral crop detection, multispectral cloud screening, and change-detection problems. Experimental results show the suitability of the proposed techniques, particularly in cases with few or poorly representative labeled samples.

Morphological attribute profiles (APs) are defined as a generalization of the recently proposed morphological profiles (MPs). APs provide a multilevel characterization of an image created by the sequential application of morphological attribute filters that can be used to model different kinds of the structural information. According to the type of the attributes considered in the morphological attribute transformation, different parametric features can be modeled. The generation of APs, thanks to an efficient implementation, strongly reduces the computational load required for the computation of conventional MPs. Moreover, the characterization of the image with different attributes leads to a more complete description of the scene and to a more accurate modeling of the spatial information than with the use of conventional morphological filters based on a predefined structuring element. Here, the features extracted by the proposed operators were used for the classification of two very high resolution panchromatic images acquired by Quickbird on the city of Trento, Italy. The experimental analysis proved the usefulness of APs in modeling the spatial information present in the images. The classification maps obtained by considering different APs result in a better description of the scene (both in terms of thematic and geometric accuracy) than those obtained with an MP.

This paper presents a novel protocol for the accuracy assessment of the thematic maps obtained by the classification of very high resolution images. As the thematic accuracy alone is not sufficient to adequately characterize the geometrical properties of high-resolution classification maps, we propose a protocol that is based on the analysis of two families of indices: 1) the traditional thematic accuracy indices and 2) a set of novel geometric indices that model different geometric properties of the objects recognized in the map. In this context, we present a set of indices that characterize five different types of geometric errors in the classification map: 1) oversegmentation; 2) undersegmentation; 3) edge location; 4) shape distortion; and 5) fragmentation. Moreover, we propose a new approach for tuning the free parameters of supervised classifiers on the basis of a multiobjective criterion function that aims at selecting the parameter values that result in the classification map that jointly optimize thematic and geometric error indices. Experimental results obtained on QuickBird images show the effectiveness of the proposed protocol in selecting classification maps characterized by a better tradeoff between thematic and geometric accuracies than standard procedures based only on thematic accuracy measures. In addition, results obtained with support vector machine classifiers confirm the effectiveness of the proposed multiobjective technique for the selection of free-parameter values for the classification algorithm.