The inherent speckle noise in synthetic aperture radar (SAR) images limits the accuracy of SAR image change detection. As a crucial step in unsupervised change detection, existing difference map generation methods primarily utilise neighbourhood information to counteract the interference caused by speckle noise. However, pixels within the neighbourhood can themselves be affected by heterogeneous pixels and noise. Therefore, this paper proposes a difference map generation method, partial neighbourhood ratio (PNR), which relies on high-confidence homogeneous pixels within the neighbourhood for difference calculation. Specifically, under the assumption that the local neighbourhood of SAR images follows a normal distribution, we develop a method for selecting high-confidence homogeneous pixels. This method quantifies inter-neighborhood dissimilarity by leveraging the statistical features of predominantly homogeneous pixel clusters within an adaptive framework, thereby reducing the impact of noise and enhancing the accuracy of difference expression. Experimental results demonstrate the superior performance of the proposed PNR. The change detection results, obtained by applying both manual trial-and-error and dual-domain network on three SAR datasets, have validated the effectiveness of the proposed algorithm.

The utilisation of synthetic aperture radar (SAR) imagery for change detection can effectively circumvents the stringent limitations imposed by weather and lighting conditions, and is finding widespread applications in fields such as disaster monitoring and urban research. To address issues of edge blurring, severe noise interference and sample imbalance, an automated SAR change detection framework is proposed based on enhanced edge information and prototype constrained clustering. Firstly, a gradient-based neighbourhood ratio is designed to reinforce the edge information of the difference map, facilitating robust differential information representations. Subsequently, to obtain accurate samples in an unsupervised manner, we have developed prototype constrained hierarchical clustering for pre-classification. The quantity and quality of selected samples can be precisely guaranteed through the utilisation of histogram analysis and prototype constraints. In the sample learning and prediction phases, a class-balanced noise-tolerant change detection network is proposed that combines focal loss and mean absolute error loss, further tackling the sample imbalance issue, strengthening noise resistance and improving change detection accuracy. Comprehensive experimental results and analysis conducted on five benchmark datasets have validated the effectiveness and robustness of the proposed method.

Efficient and reliable co-seismic landslide mapping is essential for emergency response, facilitating rescue, and post-disaster reconstruction after an earthquake. Many earthquake-prone areas globally are vegetated, making it imperative to conduct focused research on co-seismic landslides in these areas. Remote sensing images offer significant advantages for obtaining co-seismic landslide maps due to their large area coverage and short revisit time. However, many existing methods can hardly meet the accurate, robust, and efficient needs of co-seismic landslide mapping, as they encounter difficulties in differentiating co-seismic landslides from historical landslides and face complications due to other seismic-induced changes. To address these challenges, an automatic change detection-based method for fine co-seismic landslide mapping in vegetated areas was proposed. This method enabled automated landslide mapping in areas with vegetation through change detection and ensemble learning strategy incorporating logical information and spectral features of tri-temporal remote sensing images. Three experiments were carried out through PlanetScope images covering typical areas in China and Japan. The results showed that the proposed method outperformed comparative methods, including iterated principal component analysis (ITPCA), spectral angle mapper (SAM), and deep change vector analysis (DCVA). The overall accuracy (OA) of the proposed approach exceeded 94%, and both the Kappa coefficient and F1-score surpassed 0.87. These findings confirmed the superiority of the proposed method and demonstrated its promising prospects in emergency response in the future.

An origin-destination (OD) flow is the movement of objects from an origin to a destination. Determining how the flows vary across geographic locations helps understand the mechanism of flow distributions; however, it has rarely been studied. Here, we propose a trend surface model with polynomial functions to quantify the flow distribution with coordinates in the flow space. This model assumes that an observed data-record is composed of the trend value and the residual, and is represented by the orthogonal polynomial with O and D coordinates as independent variables and flow properties as dependent variables. The simulation experiments based on the linear and quadratic models indicated that the trend surface function could reflect the increasing/decreasing variation of flows with OD locations (i.e. flow trends) in different patterns. Applying this model to a case study of taxi OD flows in the broad Central Business District of Beijing, we found that the flows exhibited a rising trend toward the southwest. The trend surface characteristics are associated with the distributions of urban functional patches, where the workplaces and residences increased toward the southwest in the study area. Notably, the spatial deviations of trend surface model can help in identifying site pairs that attract flows at a high density (e.g. commerce centers and big communities), facilitating the planning of public transportation to mitigate the congestion.

Characterising spatial information as reinforcement of spectral signatures can largely assist the performance in hyperspectral image (HSI) classification. Markov random fields (MRFs) are probabilistic image texture models, and capable of encoding contextual dependencies through charactering local conditional probabilities. As a representative standardised measure of dispersion of image probability distributions, coefficient of variation (CoV) can be a useful tool for characterising spatial heterogeneity. Their parameter derivation processes also share strong compatibility with convolutional neural networks that specifies spatial correlations in local neighbourhoods. In this work, we propose an MRF and CoV based spectral-spatial convolutional network (MRF-CoV-CNN) for HSI classification. MRF models and CoVs are characterised as measures of spatial distributions and further combined with spectral information. Then the proposed MRF-CoV-CNN takes the fused features as input and produces reliable classification results. Comprehensive experiments have been conducted on the Pavia university dataset and the Salinas dataset to evaluate the proposed method both visually and quantitatively.

Change detection is a process of identifying changed ground objects by comparing image pairs obtained at different times. Compared with the pixel-level and object-level change detection, scene-level change detection can provide the semantic changes at image level, so it is important for many applications related to change descriptions and explanations such as urban functional area change monitoring. Automatic scene-level change detection approaches do not require ground truth used for training, making them more appealing in practical applications than nonautomatic methods. However, the existing automatic scene-level change detection methods only utilize low-level and mid-level features to extract changes between bitemporal images, failing to fully exploit the deep information. This article proposed a novel automatic binary scene-level change detection approach based on deep learning to address these issues. First, the pretrained VGG-16 and change vector analysis are adopted for scene-level direct predetection to produce a scene-level pseudo-change map. Second, pixel-level classification is implemented by using decision tree, and a pixel-level to scene-level conversion strategy is designed to generate the other scene-level pseudo-change map. Third, the scene-level training samples are obtained by fusing the two pseudo-change maps. Finally, the binary scene-level change map is produced by training a novel scene change detection triplet network (SCDTN). The proposed SCDTN integrates a late-fusion subnetwork and an early fusion subnetwork, comprehensively mining the deep information in each raw image as well as the temporal correlation between two raw images. Experiments were performed on a public dataset and a new challenging dataset, and the results demonstrated the effectiveness and superiority of the proposed approach

Scene change detection identifies functional changes at the scene level. Compared with pixel-level and object-level change detection, it can provide a higher level understanding of changes on the Earth’s surface. Triple-branch networks that perform scene binary change detection and scene classification tasks simultaneously are competitive in the field of scene change detection, as they consider both single-temporal scene semantic information and cross-temporal change features. However, some problems still exist. First, the temporal change feature extraction is insufficient, and the 1-D feature vector used for scene change detection and classification is lacking in representativeness. Second, the predicted scene binary change detection and classification results are often contradictory at the network prediction stage, leading to the unsatisfactory performance of change trajectory identification. To address these issues, a novel framework that integrates a differential aggregation network (DAN) and class probability-based fusion strategy (CPFS) was proposed. The designed DAN can fully capture the temporal change features using four advanced differential fusion modules (DFMs) to aggregate the multilevel difference information. In addition, it is able to generate more representative 1-D feature vectors by adopting two novel attention-aware adaptive pooling modules (AAPMs). The developed CPFS produces the final consistent scene binary change detection and classification maps by fusing three predicted class probability vectors. The proposed method was validated on two datasets, and the results demonstrated its superiority to the comparison methods.

Coal, as a traditional energy source, has made remarkable contributions to global economic development. However, surface coal mining brings a series of eco-environmental problems. Therefore, it is crucial to obtain the distribution information of coal mines. Due to the diverse appearance of coal mines and complex background environments, it is very challenging to identify coal mines at a large scale. Exposed coal is an important indicator of coal mining. Spectral indices based on satellite images possess the advantages of simplicity and high efficiency. In this study, the Exposed Coal Index (ECI) was proposed. It enables the accurate identification of exposed coal at a large scale. The effectiveness of the ECI was investigated in four typical surface coal mine distribution regions across the world. Through spectral analysis, two key characteristics of coal spectra were discovered (i.e., the flat spectral shape in the visible to near-infrared range and the low reflectance in the near-infrared band). The ECI utilized these two features to successfully differentiate coal from various background land cover types in all study cases. The results showed that the ECI was effective in visual evaluation, separability analysis, and coal mapping, with superior performance than the three previously proposed indices. The ECI can also be perfectly applied to Landsat 8 images, demonstrating its excellent generalization capability. In addition, compared with three global mining datasets, ECI provided more comprehensive information on coal mine distribution. The proposed ECI is simple, robust, and expected to provide strong support for regional resource management and sustainable development.

As an important agriculture technique, plastic greenhouse (PG) has been widely used to increase crop yield and improve food security status in the world. The high-resolution spatial information of PG is of great significance to precise agricultural management and quantitative environmental assessment. Many studies have examined the role that remote sensing technology could play in mapping and monitoring PG coverage. However, these methods, which employ either the traditional machine learning algorithms or the deep learning models, depend on massive manually labeled samples. To address this problem, this paper proposes a new cross-scale sample transferring method to generate high-resolution samples for automated PG mapping. The proposed method aims to transfer reliable label information from Sentinel-2 images (10-m) to high-resolution images (0.2-m) in a pixel-scene-pixel-object (PSPO) transferring process. In the proposed PG mapping workflow, the low-resolution label information of PG/non-PG can be obtained from an advanced plastic greenhouse index (APGI) which is calculated in Sentinel-2 images, and then the label information is transferred to the corresponding high-resolution images using the proposed PSPO transferring method. Finally, the transferred high-resolution samples are used to train the deep semantic segmentation model and produce PG mapping results. The whole process is labor-free which requires no manually labeled samples. The experimental results on three collected datasets show that the proposed approach can automatically generate accurate and reliable high-resolution samples, and the final PG mapping results can achieve an OA (overall accuracy) of 89.52% ~ 97.65% and F1 score of 84.13% ~ 94.03%, which is comparable to the fully supervised semantic segmentation model.

Accurate cropland mapping is significant for food security and sustainable development. The existing cropland map based on remote sensing mainly focus on moderate to coarse spatial resolution, and these products are generally unsuitable for precision agriculture due to the lack of spatial details. Therefore, there is an urgent need to produce high-resolution (HR) cropland maps to meet current application demands. Recently, the typical classification workflow of HR images employs deep learning models combined with manually annotated samples, and visual interpretation of samples is usually labor-intensive and time-consuming, which is not conducive to large-scale applications. To address this problem, this paper proposes an automated HR cropland extraction solution, namely RRE (Refinement-Reclassification-Extraction), including (i) Refinement of 10 m spatial resolution cropland products, (ii) Reclassifying cropland using the refined product as sample source, and (iii) Extracting HR cropland via designed cross-scale sample transfer. The strength of the proposed framework is that it leverages existing moderate-resolution public products as prior knowledge and provides cross-scale transferable samples for HR images. The whole process does not require manual labeling of samples and is highly automated. Specifically, the experimental results in the three main grain production regions show that, the RRE framework effectively reduces the interference of road networks and ridges, and F1 scores of extracted 1 m HR cropland reaches 87.71 %~94.16 %, which is comparable to the fully supervised cropland extraction method. In addition, the 10 m reclassified cropland, produced by the intermediate process of the RRE, outperforms current cropland product of ESRI Land Cover and ESA World Cover.

In the above article[1], there is a publisher typesetting error in(10), and the correct formula is

Change detection is a research hotspot in the remote sensing field. In this letter, an unsupervised change detection method was proposed by optimizing two critical steps, i.e., the generation and analysis of difference image. First, the difference vectors of features are calculated using the simple differencing method. Some changed and unchanged pixels are generated by the majority voting on the results produced by clustering the difference vectors and then are used for the weight calculation of difference vectors. The weights are calculated by means of F-Score and considered in the weighted change vector analysis to produce a discriminative difference image. Finally, the change map is obtained by the improved Markov random field which takes the difference in the neighborhood pixel values into account. Experimental results on three data sets demonstrated that the proposed method outperformed six unsupervised change detection methods in terms of overall accuracy.

Remote sensing (RS) scene classification is challenging due to changes in the scale and direction of scenes within a category. Bilinear pooling method can extract higher-order and spatial orderless information and has been shown to achieve impressive performance on various visual tasks. However, bilinear pooled features are high dimensional, which makes them impractical for subsequent processing, especially for the convolutional neural network (CNN) models with more channels in the final convolutional layer. To alleviate this shortcoming, an improved bilinear pooling method is proposed to build the compact bilinear CNN model in this work. Specifically, a joint pooling method is proposed to reduce the high-dimensional bilinear features, and it can be embedded in a bilinear CNN architecture for end-to-end optimization. Through the experimental evaluation of three real RS scene image data sets, it is proved that the improved bilinear pooling method can obtain features with higher discriminative power than the bilinear pooling method but with lower dimensionality. In addition, it also reduces the running time of model training.

Deep convolutional networks have been the most competitive method in remote sensing scene classification. Due to the diversity and complexity of scene content, remote sensing scene classification still remains a challenging task. Recently, the second-order pooling method has attracted more interest because it can learn higher-order information and enhance the nonlinear modeling ability of the networks. However, how to effectively learn second-order features and establish the discriminative feature representation of holistic images is still an open question. In this letter, we propose a first and second-order information fusion network (FSoI-Net) that can learn the first-order and second-order features at the same time, and construct the final feature representation by fusing the two types of features. Specifically, a self-attention-based second-order pooling (SaSoP) method based on covariance matrix is proposed to extract second-order features, and a fusion loss function is developed to jointly train the model and construct the final feature representation for the classification decision. The proposed network has been thoroughly evaluated on three real remote sensing scene datasets and achieved better performance than the counterparts.

In this letter, CatBoost was first introduced and investigated for remote sensing (RS) image classification using diverse features. To improve the classification performance by fostering the effective and efficient spatial feature extraction, a new pseudo label features (PLFs) extraction method was proposed via multisize neighboring patches-based multiclustering. Experimental results on two hyperspectral and one PolSAR benchmarks showed that: 1) CatBoost is an advanced ensemble learning (EL) algorithm for classification of RS images using diverse features; 2) CatBoost has better capability of reducing the overfitting issue at large number of boosting iteration; and 3) proposed PLFs can result in compatible and even better classification results than using morphological profiles (MPs) and MPs with partial reconstruction (MPPR) spatial features.

Satellite image time series classification has become a research focus with the launch of new remote sensing sensors capable of capturing images with high spatial, spectral, and temporal resolutions. In particular, in the field of crop classification, time dimension information is particularly important. Although some advanced machine learning algorithms, such as random forests (RFs), can achieve good results, they often ignore the time series information. To make full use of temporal and spectral information in multitemporal remote sensing images, a channel attention-based temporal convolutional network (CA-TCN) is proposed in this letter. Specifically, the proposed method is composed of two main modules: temporal convolutional network and attention block. The temporal convolutional network can capture long-range dependence by using a hierarchy of temporal convolutional filters. To capture relevant information inside the sequence and enhance the important information, the attention block is used to enhance the important features in the channel dimension since not all bands contain equal information in crop type classification. The proposed CA-TCN can excavate deeper phenological characteristics. Compared to the temporal attention-based temporal convolutional network and other deep learning-based models, the proposed CA-TCN has achieved state-of-the-art performance in the Breizhcrops dataset with fewer parameters.

Continuous change detection (CCD) is one of the most famous algorithms in remote sensing time series change detection, making any improvements on it are of great significance for the practice of monitoring land cover dynamics. Inspired by the directionality of CCD, a bidirectional CCD (Bi-CCD) is proposed to improve the accuracy of change detection by selecting the optimal result from the results of CCD running from both the forward and backward directions of time series. Three criteria including root mean square error (RMSE), Akaike information criterion (AIC), and Bayesian information criterion (BIC) were adopted to define the “optimal” in this study, and their effects under three common parameter settings were evaluated in a simulation dataset. The quantitative results show that Bi-CCD based on different result selection criteria can indeed improve the accuracy of change detection in terms of omission error and commission error. In general, Bi-CCD based on RMSE achieved the lowest omission error, while Bi-CCD based on BIC obtained the lowest commission error. Compared with the unidirectional CCD, Bi-CCD reduces the omission error and commission error by at most 11.19% and 15.08%, respectively. In addition, the different effects of different optimal result selection criteria on the accuracy of Bi-CCD enable Bi-CCD more flexible to handle different tasks than the standard CCD.

The ZY1-02D satellite, which was launched in 2019, is China’s first civil hyperspectral satellite. However, the laboratory calibration and vicarious calibration methods could not provide accurate radiometric calibration coefficients after the satellite had been launched. In this article, we describe how a cross-calibration method was utilized to calibrate the ZY1-02D hyperspectral imager using the well-calibrated Gaofen-5 Advanced Hyperspectral Imager (GF-5 AHSI). The 6S radiative transfer model was selected to simulate the apparent reflectance of the two hyperspectral sensors under corresponding imaging conditions, and the calibration coefficients were calculated by spectral channel matching. The reflectance-based vicarious calibration was carried out for comparison. Through the validation experiments, it is shown that the reflectance data obtained by cross-calibration and vicarious calibration are basically consistent, showing a stable radiation performance. At the Dunhuang calibration site, the ratio of measured surface reflectance to the cross-calibrated image reflectance is between 0.9 and 1.1, the$R^{2}$values are more than 0.96, and the spectral angles are less than 3°. The validation results for different ground features also show the applicability of the corrected coefficients. When compared with different sensors, the maximum difference between the ZY1-02D reflectance results after cross-calibration and Landsat-8/Sentinel-2 is less than 0.04 and the mean difference is less than 0.02, which further proves that the ZY1-02D hyperspectral imager has a high radiation accuracy after cross-calibration. The proposed cross-calibration method could be used as an effective supplement to the on-orbit calibration method and could also be extended to other satellite hyperspectral imagers.

The highly correlated spectral features and the limited training samples pose challenges in hyperspectral image classification. In this article, to tackle the issues of end-to-end feature learning and transfer learning with limited labeled samples, we propose a unified multiscale learning (UML) framework, which is based on a fully convolutional network. A multiscale spatial-channel attention mechanism and a multiscale shuffle block are proposed in the UML framework to improve the problem of land-cover map distortion. The contextual information and the spectral feature are enhanced before the last classification layer based on three strategies in this work: 1) the channel shuffle operation, which was employed to learn the more effective spectral characteristics by disordering the channels of the feature map; 2) multiscale block, which considered the contextual information in multiple ranges; and 3) spatiospectral attention, which enhanced the expression of the important characteristic among all pixels. Three hyperspectral datasets, including two airborne hyperspectral images and one spaceborne hyperspectral image, were used to demonstrate the performance of the UML framework in both classification and transfer learning. The experimental results confirmed that the proposed method outperforms most of the state-of-the-art hyperspectral image classification methods. The source code is released athttps://github.com/Hyper-NN/UML.

Deep learning (DL) has shown great potentials for hyperspectral image (HSI) classification due to its powerful ability of nonlinear modeling and end-to-end optimization. However, DL models are easily get trapped into overfitting due to limited training labels since the labeling process is time-consuming and laborious in real classification scenario. To overcome this issue, we propose a novel spectral-spatial siamese network (S3Net) for few-shot HSI classification. Firstly, a lightweight spectral-spatial network (SSN) composed of 1-D and 2-D convolution is proposed to extract spectral-spatial features. Secondly, S3Net is constructed by two SSNs in dual branches, which can augment training set by feeding sample pairs into each branch, and thus enhancing the model separability. To provide more features for the model, differentiated patches are fed into each branch, where negative samples are random selected to avoid redundancy. Finally, a weighted contrastive loss is designed to promote the model to fit in the right direction by focusing on sample pairs that are hardly to be identified. Moreover, another adaptive cross entropy loss is conceived to learn the fusion ratio of the two branches. Experiments based on three commonly used HSI data sets demonstrate that S3Net outperforms traditional and state-of-the-art DL-based HSI classification methods under few-shot training scenario. In addition, the weighted contrastive loss and the adaptive cross entropy loss jointly improve the discrimination power of the model.

An effective network structure is essential for the classification of satellite image time series (SITS). Deep learning models have been widely used for SITS classification and achieved impressive performance, especially the architectures based on self-attention. However, the lack of efficient and comprehensive attention to valuable bands and time series structure hinders the performance to some extent. To address this problem, an end-to-end attention-aware dynamic self-aggregation network (ADSN) is proposed for SITS classification in this work, which combines two main parts: spectral focusing and spectral–temporal feature learning. The core components of ADSN are the channel attention module and dynamic self-aggregation block. Specifically, informative bands in the SITS flowing through the channel attention module can adaptively get a high weight to increase their contributions, while the attentions of some low-efficiency bands are weakened. Besides, the dynamic self-aggregation block, which integrates multiscale dynamic convolution and improved multihead attention in parallel, can simultaneously capture long- and short-distance sequence structures and position relationships to better represent temporal information. Compared with random forest (RF) and seven deep learning algorithms, the proposed model effectively learns spectral and temporal features, and the experimental results confirm that ADSN has achieved superior classification accuracy and generalization ability on two SITS datasets with extremely unbalanced samples.

The visible-shortwave infrared Advanced Hyperspectral Imager (AHSI) is a payload onboard the Gaofen-5 satellite, which is China’s first hyperspectral satellite and is part of the Chinese High-Resolution Earth Observation System. As a supplement to the onboard radiometric calibration of the AHSI instrument, vicarious calibration is also required, which is independent of the instrument-based calibration. In this article, a reflectance-based vicarious calibration approach is presented, which takes surface reflectance data, aerosol data, and atmospheric water vapor data into account. The Dunhuang test site, which is one of the China Radiometric Calibration Sites (CRCS) for the vicarious calibration of spaceborne sensors, possesses stable, uniform, and measurable surface objects, so it was chosen as the radiation source to replace the laboratory and onboard calibrators. A Spectra Vista Corporation (SVC) spectral radiometer and a CE318 sun photometer were utilized for the measurement of the surface reflectance and the condition of the aerosol, respectively. The radiance at the entrance pupil at the top of atmosphere was then obtained through the MODerate resolution atmospheric TRANsmission (MODTRAN) atmospheric transmission model. The surface reflectance was obtained using the Fast Line-of-sight Atmospheric Analysis of Hypercubes (FLAASH) atmospheric model for validation. The results show that, with regard to the calibration coefficients, the calibrated AHSI instrument presents a stable radiometric performance among different land-cover types. The ratios on all the bands are between 0.8 and 1.2 and are consistent with the reflectance data from the Dunhuang test site. The${R} ^{{2}}$values are all greater than 0.95 and the spectral angle is all less than 2°. The standard deviations of the ratios are less than 3% for each chosen band, which proves that the calibrated data have a high consistency with thein situmeasurements. When compared with Landsat 8 and Sentinel-2, the mean errors of the surface reflectance are all under 0.06, which further demonstrates that the calibrated reflectance has a high accuracy.

Deep learning (DL) has exhibited huge potentials for hyperspectral image (HSI) classification due to its powerful nonlinear modeling and end-to-end optimization characteristics. Although the superior performance of DL-based methods has been witnessed, some limitations can still be found. On the one hand, existing DL frameworks usually resorted to first-order statistical features, whereas they rarely considered second-order or higher order statistical features. On the other hand, the optimization of complex hyperparameters (e.g., the layer number and convolutional kernel size) is time-consuming and a very tough task, making the designed DL framework unexplainable. To overcome these challenges, we propose a novel attention-based second-order pooling network (A-SPN). First, a first-order feature operator is designed to model the spectral–spatial information of HSI. Second, an attention-based second-order pooling (A-SOP) operator is designed to model discriminative and representative features. Finally, a fully connected layer with softmax loss is used for classification. The proposed framework can obtain second-order statistical features in an end-to-end manner. In addition, A-SPN is free of complex hyperparameters tuning, making it more explainable and easily equipped for classification tasks. Experimental results based on three common hyperspectral data sets demonstrate that A-SPN outperforms other traditional and state-of-the-art DL-based HSI classification methods in terms of generalization performance with limited training samples, classification accuracy, convergence rate, and computational complexity.

To fuse hyperspectral and Light Detection And Ranging (LiDAR), we propose a semisupervised graph fusion (SSGF) approach. We apply morphological filters to LiDAR and the first few components of hyperspectral data to model the height and spatial information, respectively. Then, the proposed SSGF is used to project the spectral, elevation, and spatial features onto a lower subspace to obtain the new features. In particular, the objective of SSGF is to maximize the class separation ability and preserve the local neighborhood structure by using both labeled and unlabeled samples. Experimental results on the hyperspectral and LiDAR data from the 2013 IEEE Geoscience and Remote Sensing Society (GRSS) Data Fusion Contest demonstrated the superiority of the SSGF.

Recently, collaborative representation classification (CRC) has attracted much attention for hyperspectral image analysis. In particular, tangent space CRC (TCRC) has achieved excellent performance for hyperspectral image classification in a simplified tangent space. In this article, novel Bagging-based TCRC (TCRC-bagging) and Boosting-based TCRC (TCRC-boosting) methods are proposed. The main idea of TCRC-bagging is to generate diverse TCRC classification results using the bootstrap sample method, which can enhance the accuracy and diversity of a single classifier simultaneously. For TCRC-boosting, it can provide the most informative training samples by changing their distributions dynamically for each base TCRC learner. The effectiveness of the proposed methods is validated using three real hyperspectral data sets. The experimental results show that both TCRC-bagging and TCRC-boosting outperform their single classifier counterpart. In particular, the TCRC-boosting provides superior performance compared with the TCRC-bagging.

Deep generative models such as the generative adversarial network (GAN) and the variational autoencoder (VAE) have obtained increasing attention in a wide variety of applications. Nevertheless, the existing methods cannot fully consider the inherent features of the spectral information, which leads to the applications being of low practical performance. In this article, in order to better handle this problem, a novel generative model named the conditional variational autoencoder with an adversarial training process (CVA2E) is proposed for hyperspectral imagery classification by combining variational inference and an adversarial training process in the spectral sample generation. Moreover, two penalty terms are added to promote the diversity and optimize the spectral shape features of the generated samples. The performance on three different real hyperspectral data sets confirms the superiority of the proposed method.

This study tested and compared the suitability of SPOT-5 image and LiDAR data both separately and combined for the classification of the urban area using Rotation Forest (ROF) classifier. Experimental results revealed that the integration of the SPOT-5 image and LiDAR data classification scheme gave better classification accuracies, when compared to the classification schedules using one of the two solely. Furthermore, RoF classifier produced better classification results than that of other two classifiers (i.e., SVMs (Support Vector Machines) and RF (Random Forests)). Finally, it should be noted that RoF classifier provided an effective way of combining SPOT-5 image and LiDAR data for classification, which is robust to the important parameter M (i.e. the number of features in each subset).

Spatial information has widely been used in hyperspectral image (HSI) classification to improve classification accuracy. However, the structural information may not be fully explored when using spatial information, this paper proposes the joint collaborative representation classification with correlation matrix (CRC-CM) for HSI by using spatial correlation features in patches, which could keep the local intrinsic structure in band images. Considering spatial heterogeneity in a patch, local correlation matrices of a target neighborhood patch and training neighborhood patch are improved by a binary weight matrix and shape-adaptive neighborhood. To explore nonlinear nature of spatial features, corresponding kernel CRC-CM is also proposed. To evaluate the effectiveness of the proposed methods, three real HSIs with different degree of heterogeneity are used. The experimental results show that the proposed spatial correlation features outperform the original spectral feature and other spatial features which widely used in HSI classifiers.

The increase in the spectral and spatial information of hyperspectral imagery poses challenges in classification due to the fact that spectral bands are highly correlated, training samples may be limited, and high resolution may increase intraclass difference and interclass similarity. In this paper, in order to better handle these problems, a Caps-TripleGAN framework is proposed by exploring the 1-D structure triple generative adversarial network (TripleGAN) for sample generation and integrating CapsNet for hyperspectral image classification. Moreover, spatial information is utilized to verify the learning capacity and discriminative ability of the Caps-TripleGAN framework. The experimental results obtained with three real hyperspectral data sets confirm that the proposed method outperforms most of the state-of-the-art methods.

Recently, deep learning (DL) is of great interest in hyperspectral image (HSI) classification. Although many effective frameworks exist in the literature, the generally limited availability of training samples poses great challenges in applying DL to HSI classification. In this paper, we present a novel DL framework, namely, semisupervised stacked autoencoders (Semi-SAEs) with cotraining, for HSI classification. First, two SAEs are pretrained based on the hyperspectral features and the spatial features, respectively. Second, fine-tuning is alternatively conducted for the two SAEs in a semisupervised cotraining fashion, where the initial training set is enlarged by designing an effective region growing method. Finally, the classification probabilities obtained by the two SAEs are fused using a Markov random field model solved by iterated conditional modes. Experimental results based on three popular hyperspectral data sets demonstrate that the proposed method outperforms other state-of-the-art DL methods.

In this letter, an unsupervised 2-D and 3-D urban change detection scheme is proposed exploiting Quad-PolSAR data. Changes are extracted by segmenting the data into superpixels, to enhance the balance among change components and increase estimability of prior distributions. Positive and negative change components for built-up areas, in both the horizontal and the vertical directions, are properly extracted by assuming a multivariate Gaussian mixed model applied to a subset of polarimetric parameters at the superpixel level. The proposed method is tested on multitemporal Quad-PolSAR images and the results confirm its effectiveness. The selection of polarimetric decomposition measures that are most useful to the task is also experimentally justified.

In this letter, multisource earth observation (EO) data sets, including multitemporal Landsat-8, digital surface model, and spatial information, were integrated for land-cover classification by random forest (RF) and support vector machines (SVMs). We demonstrated in this letter that both RF and SVM are useful tools for classification of land cover in the local climate zones featured with highly heterogeneous landscape. Classification of land cover by RF was with an overall accuracy (OA) of 86.2%, while the OA was 85.5% for SVM. However, we found that RF was more stable than SVM for multisource EO data in classifying land cover without normalizing different feature data sets. Experiments showed that the thermal features were more important than temporal and spatial ones in discriminating impervious objects, while the temporal and spatial features were generally better than thermal ones in separating the distinct vegetation categories. Another finding was that our experiments indicated that spectral features were the most important in classification of land cover, followed by temporal, thermal, and spatial features, respectively. As to the spectral features, red channels were the most important, followed by short-wave infrared, near-infrared, and green channels. Thus, it could be concluded that the combination of spectral, thermal, spatial, and temporal information would be an optimal approach to increase the OA of land-cover classification in the zones featured with highly heterogeneous landscape.

To adequately represent the nonlinearities in the high-dimensional feature space for hyperspectral images (HSIs), we propose a multiple kernel collaborative representation-based classifier (CRC) in this paper. Extended morphological profiles are first extracted from the original HSIs, because they can efficiently capture the spatial and spectral information. In the proposed method, a novel multiple kernel learning (MKL) model is embedded into CRC. Multiple kernel patterns, e.g., Naive, Multimetric, and Multiscale are adopted for the optimal set of basic kernels, which are helpful to capture the useful information from different pixel distributions, kernel metric spaces, and kernel scales. To learn an optimal linear combination of the predefined basic kernels, we add an extra training stage to the typical CRC where kernel weights are jointly learned with the representation coefficients from the training samples by minimizing the representation error. Moreover, by considering different contributions of dictionary atoms, the adaptive representation strategy is applied to the MKL framework via a dissimilarity-weighted regularizer to obtain a more robust representation of test pixels in the fused kernel space. Experimental results on three real HSIs confirm that the proposed classifiers outperform the other state-of-the-art representation-based classifiers.

Multiple types of features, e.g., spectral, filtering, texture, and shape features, are helpful for hyperspectral image (HSI) classification tasks. Combining multiple features can describe the characteristics of pixels from different perspectives, and always results in better classification performance. Recently, multifeature combination learning has been widely employed to the multitask-learning-based representation-based model to obtain a multifeature representation vector. However, the linear sparse representation-based classifier (SRC) cannot handle the HSI with highly nonlinear distribution, and kernel sparse representation-based classifier (KSRC) can remedy the drawback of linear SRC. By adopting nonlinear mapping, the samples in kernel space are often of high or even infinite dimensionality. In this paper, we integrate kernel principal component analysis into multifeature-based KSRC and propose a novel multiple feature kernel sparse representation-based classifier (namely, MFKSRC) for hyperspectral imagery. More specifically, spatial features, Gabor textures, local binary patterns, and difference morphological profiles are adopted and then each kind of feature is transformed nonlinearly into a new low-dimensional kernel space. The proposed framework can handle data with nonlinear distribution and add a dimensionality reduction stage in kernel space before optimizing the corresponding cost function. Experimental results on different HSIs demonstrate that the proposed MFKSRC algorithm outperforms the state-of-the-art classifiers.

Superpixels are a powerful device to characterize the spatial-contextual information in remotely sensed hyperspectral image (HSI) interpretation. However, the exploitation of superpixels in classification problems is not straightforward, often leading to unbearable NP-hard discrete integer optimization problems. In this paper, we attack this hurdle by leveraging on a convex relaxation of the original integer optimization problem, which opens the door to include oversegmented superpixel-based regularizers. Specifically, we develop a new method for generating oversegmented superpixels. Then, we introduce a family of convex regularizers in the form of graph total variation, which promotes the same labeling in each superpixel. Vectorial total variation is also included in order to promote piecewise smoothness and align discontinuities along the class boundaries. The solution of the obtained convex optimization problem is computed with the split-augmented Lagrangian shrinkage algorithm. Experiments on HSIs yield classification maps with precise boundaries and inner consistency inside oversegmented superpixels, leading to the state-of-the-art classification accuracies.

Recently, multifeature learning in collaborative representation classification (CRC) for hyperspectral images has generated promising performance. In this paper, two novel multifeature learning algorithms that update dictionary directly and indirectly are proposed. In order to offer the complementarity of multifeature, four different types of features-global feature (i.e., Gabor feature), local feature (i.e., local binary pattern), shape feature (i.e., extended multiattribute profiles), and spectral feature-are adopted in this paper. Under the hypothesis that most of the features should share the same coding pattern in CRC, this paper proposes to learn proper dictionaries for each feature until obtaining stable codes in a linear classifier. Furthermore, to avoid the explicit mapping of infinite-dimensional dictionaries in a nonlinear kernelized classifier, an indirect approach to construct the transformation matrix from original dictionaries to learn new dictionaries is developed. Three real hyperspectral images acquired from different sensors are adopted for performance evaluation. The experimental results demonstrate that the proposed methods can provide superior performance compared with those of the state-of-the-art classifiers.

In this work, we develop a new framework to combine ensemble learning and composite kernel learning for hyperspectral image classification. We refer it as the multiple composite kernel learning, which is based on an iterative architecture. More specifically, in each iteration, we use the rotation-based ensemble to create rotation matrix, which is used to generate rotated features for both spectral and spatial information (e.g., extinction profiles). Then, the new spectral and spatial features are integrated into the composite kernels based on support vector machines classifier. Different rotation matrices will lead to obtaining various newly spectral and spatial characteristics, thereby they further increase the diversity and the classification performance. Experimental results on Indian Pines benchmark hyperspectral dataset demonstrate the excellent performance of the proposed method.

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.

In this letter, we propose a kernel fused representation-based classifier (KFRC) for hyperspectral images (HSIs), which combines sparse representation (SR) and collaborative representation (CR) into a unified kernel representation-based classification framework. First, we present two individual kernel methods, i.e., kernel SR (KSR) and kernel CR (KCR), which kernelize the representation methods by projecting the samples into a high-dimensional kernel space to improve the samples separability between different classes. Once obtaining the two kernel representation coefficients, KFRC attempts to achieve a balance between KSR and KCR via an adjusting parameter $\theta $ in the kernel residual domain. Subsequently, the class label of each test sample is determined by the minimum residual for each class. Experimental results on two HSIs demonstrate the proposed kernel fused method performs better than the other state-of-the-art representation-based classifiers.

To improve the capability of a traditional sparse representation-based classifier (SRC), we propose a novel dissimilarity-weighted SRC (DWSRC) for hyperspectral image (HSI) classification. In particular, DWSRC computes the weights for each atom according to the distance or dissimilarity information between the test pixel and the atoms. First, a locality constraint dictionary set is constructed by the Gaussian kernel distance with a suitable distance metric (e.g., Euclidean distance). Second, the test pixel is sparsely coded over the new weighted dictionary set based on the 11-norm minimization problem. Finally, the test pixel is classified by using the obtained sparse coefficients with the minimal residual rule. Experimental results on two widely used public HSIs demonstrate that the proposed DWSRC is more efficient and accurate than other state-of-the-art SRCs.

Scene classification from remote sensing images provides new possibilities for potential application of high spatial resolution imagery. How to efficiently implement scene recognition from high spatial resolution imagery remains a significant challenge in the remote sensing domain. Recently, convolutional neural networks (CNN) have attracted tremendous attention because of their excellent performance in different fields. However, most works focus on fully training a new deep CNN model for the target problems without considering the limited data and time-consuming issues. To alleviate the aforementioned drawbacks, some works have attempted to use the pretrained CNN models as feature extractors to build a feature representation of scene images for classification and achieved successful applications including remote sensing scene classification. However, existing works pay little attention to exploring the benefits of multilayer features for improving the scene classification in different aspects. As a matter of fact, the information hidden in different layers has great potential for improving feature discrimination capacity. Therefore, this paper presents a fusion strategy for integrating multilayer features of a pretrained CNN model for scene classification. Specifically, the pretrained CNN model is used as a feature extractor to extract deep features of different convolutional and fully connected layers; then, a multiscale improved Fisher kernel coding method is proposed to build a mid-level feature representation of convolutional deep features. Finally, the mid-level features extracted from convolutional layers and the features of fully connected layers are fused by a principal component analysis/spectral regression kernel discriminant analysis method for classification. For validation and comparison purposes, the proposed approach is evaluated via experiments with two challenging high-resolution remote sensing data sets, and shows the competitive performance compared with fully trained CNN models, fine-tuning CNN models, and other related works.

Regularization has appeared explicitly in hyperspectral image (HSI) classification community, which serves as a promising paradigm for leveraging labeled and unlabeled information, computer's automation and user's interaction, spectral and spatial information, and so on. Graph-based regularization is capable of modeling the nonlinear structures embedded in high-dimensional space, with the great potential for HSI classification. However, traditional methods exhibit low capacity when facing noisy and large-scale data, thus posing a big challenge for their successful use in this community. In this paper, we present two novel sparse graph regularization methods, SGR and SGR with total variation (TV-SGR). In SGR, the labels of large unknown data are propagated based on the fraction matrix and the prediction function, where the fraction matrix is obtained using an effective sparse representation (SR) algorithm with respect to the dictionary, and the prediction function is estimated by optimizing a typical graph-based regularization problem. In contrast, TV-SGR is an extension of SGR by considering spatial information modeled by total variation in SR. Propagating the prediction function from dictionary to large unknown data using the fraction matrix is the essence of the paradigm. SGR and TV-SGR can be equipped with semisupervised learning, active learning, and spectral-spatial classification with large flexibility. The experimental results with two popular hyperspectral data sets indicate that the proposed methods outperform some state-of-the-art approaches in terms of computational efficacy, classification accuracy, and robustness to noise.

The potential hazard of heavy metals in reclaimed mine soil has been attracted more and more attention. Hyperspectral inversion can be applied to predict the heavy metal content of the soil effectively. Three machine learning methods, Support Vector Machine (SVM), Random Forest (RF) and Extreme Learning Machine (ELM), are introduced in this paper, and then are compared with the Partial Least Squares (PLS) method. With the correlation analysis of heavy metal content and pretreatment spectral band, the models are constructed to predict the content of heavy metal in soil. The results show that the prediction results of machine learning methods are better than PLS, and ELM and RF are better than SVM. Analyzing the stability of the model, it can be found that the concentration of heavy metal samples will affect the prediction of ELM. Meanwhile, the stability of RF is the best than the other three models. RF algorithm has also the highest accuracy in the inversion of soil heavy metal research.

Rotation Forest (RoF) is a decision tree ensemble classifier, which uses random feature selection and data transformation techniques to improve both the diversity and accuracy of base classifiers. Traditional RoF only considers data transformation on spectral information. In order to further improve the performance of RoF, we introduce spectral-spatial data transformation into RoF and thus propose a spectral-spatial Rotation Forest (SSRoF). The proposed method is experimentally investigated on a hyperspectral remote sensing image collected by the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor. Experimental results indicate that the proposed methodology achieves excellent performance.

In this research, a novel sparse graph regularization (SGR) method was presented, aiming at robust crop mapping using hyperspectral imagery with very few in situ data. The core of SGR lies in propagating labels from known data to unknown, which is triggered by: 1) the fraction matrix generated for the large unknown data by using an effective sparse representation algorithm with respect to the few training data serving as the dictionary; 2) the prediction function estimated for the few training data by formulating a regularization model based on sparse graph. Then, the labels of large unknown data can be obtained by maximizing the posterior probability distribution based on the two ingredients. The study area is located at Zhangye oasis in the middle reaches of Heihe watershed, Gansu, China, where eight crop types were mapped with Compact Airborne Spectrographic Imager (CASI) and Shortwave Infrared Airborne Spectrogrpahic Imager (SASI) hyperspectral data. Experimental results demonstrate that the proposed method significantly outperforms other classifiers, with an overall accuracy of 87.43% and a kappa value of 0.827 (5 labeled samples per class), which are respectively, 9%-30% and 0.1-0.3 higher than other counterparts.

Morphological attribute profiles (APs) are discriminant features in the spectral–spatial classification of hyperspectral data. However, the optimal range of parameters in each filter is always a challenging yet important task, since an unsuitable range of parameters likely leads to inferior results. In order to alleviate this problem, we propose an ensemble method, which integrates multiple classification results based on a series of APs. The APs are obtained by using different filters with thresholds that are randomly selected from an arbitrarily defined range of parameters. Experimental results conducted on two hyperspectral images demonstrate the robustness and effectiveness of the proposed method.

In this letter, we propose a new version of the rotation forest (RoF) method for the pixelwise classification of hyperspectral images. RoF, which is an ensemble of decision tree classifiers, uses random feature selection and data transformation techniques (i.e., principal component analysis) to improve both the accuracy of base classifiers and the diversity within the ensemble. Traditional RoF performs data transformation on the training samples of each subset. In order to further improve the performance of RoF, the data transformation is separately performed on each class, extracting sets of transformation matrices that are strictly dependent on the training samples of each single class. The approach, namely, class-separation-based RoF (RoFCS), is experimentally investigated on a hyperspectral image collected by Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor. Experimental results demonstrate that the proposed methodology achieves excellent performances, in comparison with random forest and RoF classifiers.

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.

Active learning (AL) methods that select unlabeled samples only querying by informative measures (i.e., uncertainty and/or diversity criteria) have been extensively investigated. However, these methods usually do not exploit the manifold structure of the unlabeled data from the geometrical point of view, a choice that might lead to a sample bias and consequently undesirable performances. To control and possibly overcome such drawbacks, this paper explores AL methods based on joint informative and manifold structure representative sampling (JI-MSRS). In JI-MSRS, a portion of the unlabeled samples that are added at each iteration is selected according to the informative measures, whereas another portion is selected according to their capability to represent the data cluster structure. Four popular manifold learning methods, namely, principle component analysis (PCA), linear discriminant analysis, kernel PCA, and neighborhood preserving embedding, are used to model the data structure. Then, Delaunay triangulation nets are used to build a discrete approximation of the geometrical structure of the unlabeled data cloud in a low-dimensional space. To show the effectiveness of this novel sampling strategy, results on three real multi-/hyperspectral data sets are presented, adding a thorough comparison with other state-of-the-art AL techniques. In comparison to conventional AL heuristics, the proposed techniques are able to obtain competitive or even better classification accuracy values.

With different principles, support vector machines (SVMs) and multiple classifier systems (MCSs) have shown excellent performances for classifying hyperspectral remote sensing images. In order to further improve the performance, we propose a novel ensemble approach, namely, rotation-based SVM (RoSVM), which combines SVMs and MCSs together. The basic idea of RoSVM is to generate diverse SVM classification results using random feature selection and data transformation, which can enhance both individual accuracy and diversity within the ensemble simultaneously. Two simple data transformation methods, i.e., principal component analysis and random projection, are introduced into RoSVM. An empirical study on three hyperspectral data sets demonstrates that the proposed RoSVM ensemble method outperforms the single SVM and random subspace SVM. The impacts of the parameters on the overall accuracy of RoSVM (different training sets, ensemble sizes, and numbers of features in the subset) are also investigated in this paper.

This paper proposes a semi-supervised graph-based fusion framework to couple dimensionality reduction and the fusion of multi-sensor data for classification. First, morphological features are used to model the elevation and spatial information contained in both LiDAR data and on the first few principal components (PCs) of the original hyperspectral (HS) image. Then, we fuse the features by projecting the spectral, spatial and elevation features onto a lower subspace through our proposed semi-supervised fusion graph. Experimental results on fusion of HS and LiDAR data from the 2013 IEEE GRSS Data Fusion Contest demonstrate effectiveness of the proposed method. Compared to the methods using single data source or unsupervised graph fusion, with the proposed method, overall classification accuracies were improved by 9% and 4%, respectively.

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.

In this letter, we propose a novel approach for improving Random Forest (RF) in hyperspectral image classification. The proposed approach combines the ensemble of features and the semisupervised feature extraction (SSFE) technique. The main contribution of our approach is to construct an ensemble of RF classifiers. In this way, the feature space is divided into several disjoint feature subspaces. Then, the feature subspaces induced by the SSFE technique are used as the input space to an RF classifier. This method is compared with a regular RF and an RF with the reduced features by the SSFE on two real hyperspectral data sets, showing an improved performance in ill-posed, poor-posed, and well-posed conditions. An additional study shows that the proposed method is less sensitive to the parameters.

The quantitative estimation of the fractional cover of carbonate rock (CR) is critical for natural resource management and ecological conservation in karst areas. Based on the analysis of spectral properties of CR together with other land cover types, we proposed two CR indices (CRIs) and established the model that represents the relationships between the CRIs and the fractional cover of CR. Then, the fractional cover of CR was estimated by using the developed model. Experimental results on Landsat-8 Operational Land Imager images acquired at Southwestern China demonstrated the effectiveness of the developed model. Compared with other indices, the proposed CRIs show the highest correlations with the fractional cover of CR.

Accurate and rapid mapping of shallow water bathymetry is essential for the safe operation of many industries. Here, we propose a new approach to shallow water bathymetry mapping that integrates hyperspectral image and sparse sonar data. Our approach includes two main steps: dimensional reduction of Hyperion images and interpolation of sparse sonar data. First, we propose a new algorithm, i.e., a sonar-based semisupervised Laplacian eigenmap (LE) using both spatial and spectral distance, for dimensional reduction of Hyperion imagery. Second, we develop a new algorithm to interpolate sparse sonar points using a 3-D information diffusion method with homogeneous regions. These homogeneous regions are derived from the segmentation of the dimensional reduction results based on depth. We conduct the experimental comparison to confirm the applicability of the dimensional reduction and interpolation methods and their advantages over previously described methods. The proposed dimensional reduction method achieves better dimensional results than unsupervised method and semisupervised LE method (using only spectral distance). Furthermore, the bathymetry retrieved using the proposed method is more precise than that retrieved using common interpolation methods.

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.

In this paper, we propose a new spectral-spatial classification strategy to enhance the classification performances obtained on hyperspectral images by integrating rotation forests and Markov random fields (MRFs). First, rotation forests are performed to obtain the class probabilities based on spectral information. Rotation forests create diverse base learners using feature extraction and subset features. The feature set is randomly divided into several disjoint subsets; then, feature extraction is performed separately on each subset, and a new set of linear extracted features is obtained. The base learner is trained with this set. An ensemble of classifiers is constructed by repeating these steps several times. The weak classifier of hyperspectral data, classification and regression tree (CART), is selected as the base classifier because it is unstable, fast, and sensitive to rotations of the axes. In this case, small changes in the training data of CART lead to a large change in the results, generating high diversity within the ensemble. Four feature extraction methods, including principal component analysis (PCA), neighborhood preserving embedding (NPE), linear local tangent space alignment (LLTSA), and linearity preserving projection (LPP), are used in rotation forests. Second, spatial contextual information, which is modeled by MRF prior, is used to refine the classification results obtained from the rotation forests by solving a maximum a posteriori problem using the α-expansion graph cuts optimization method. Experimental results, conducted on three hyperspectral data with different resolutions and different contexts, reveal that rotation forest ensembles are competitive with other strong supervised classification methods, such as support vector machines. Rotation forests with local feature extraction methods, including NPE, LLTSA, and LPP, can lead to higher classification accuracies than that achieved by PCA. With the help of MRF, the proposed algorithms can improve the classification accuracies significantly, confirming the importance of spatial contextual information in hyperspectral spectral-spatial classification.

Classification is one of the most important techniques to the analysis of hyperspectral remote sensing images. Nonetheless, there are many challenging problems arising in this task. Two common issues are the curse of dimensionality and the spatial information modeling. In this paper, we present a new general framework to train series of effective classifiers with spatial information for classifying hyperspectral data. The proposed framework is based on the two key observations: 1) the curse of dimensionality and the high feature-to-instance ratio can be alleviated by using random subspace (RS) ensembles; and 2) the spatial-contextual information is modeled by the extended multiattribute profiles (EMAPs). Two fast learning algorithms, i.e., decision tree (DT) and extreme learning machine (ELM), are selected as the base classifiers. Six RS ensemble methods, namely, RS with DT, random forest (RF), rotation forest, rotation RF (RoRF), RS with ELM (RSELM), and rotation subspace with ELM (RoELM), are constructed by the multiple base learners. Experimental results on both simulated and real hyperspectral data verify the effectiveness of the RS ensemble methods for the classification of both spectral and spatial information (EMAPs). On the University of Pavia Reflective Optics Spectrographic Imaging System image, our proposed approaches, i.e., both RSELM and RoELM with EMAPs, achieve the state-of-the-art performances, which demonstrates the advantage of the proposed methods. The key parameters in RS ensembles and the computational complexity are also investigated in this paper.

Sparse graph embedding (SGE) is a promising technique useful for the nonlinear feature extraction (FE) of hyperspectral images (HSIs). However, such images exhibit spatial variability and spectral multimodality, presenting challenges to existing FE methods, including SGE. To address this issue, this paper presents two novel SGE methods for HSI classification. One method, which is termed simultaneous SGE (SSGE), is designed to consider the spatial variability of spectral signatures by using a simultaneous sparse representation (SSR) model integrated with a shape-adaptive neighborhood building approach. In addition, a sparse graph is constructed via matrix computation based on sparse codes. Then, low-dimensional features are produced by employing linear graph embedding (LGE) based on the constructed sparse graph. The other method, which is termed simultaneous sparse multimanifold learning (SSMML), is proposed to handle the multimodality of an HSI. In SSMML, multiple views are generated to represent different modalities. Then, multiview-oriented submanifolds are produced by adopting SSGE, and they are further integrated via coregularization. SSGE is capable of modeling both local and global data structures. Furthermore, SSMML serves as a prototype that can model multimodal data structures. The proposed methods are evaluated by using sparse multinomial logistic regression for HSI classification. Experimental results with two popular hyperspectral data sets validate the good performance of the two methods in producing more representative low-dimensional features and yielding superior classification results compared with other related approaches.

This paper presents a new spectral-spatial classification method for hyperspectral images via morphological component analysis-based image separation rationale in sparse representation. The method consists of three main steps. First, the high-dimensional spectral domain of hyperspectral images is reduced into a low-dimensional feature domain by using minimum noise fraction (MNF). Second, the proposed separation method is acted on each features to generate the morphological components (MCs), i.e., the content and texture components. To this end, the dictionaries for these two components are built by using local curvelet and Gabor wavelet transforms within the randomly chosen image partitions. Then, sparse coding of one of the MCs and update of the associated dictionary are sequentially performed with the other one fixed. To better direct the separation process, an undecimated Haar wavelet with soft threshold is performed for the content component to make it smooth. This process is repeated until some stopping criterion is met. Finally, a support vector machine is adopted to obtain the classification maps based on the MCs. The experimental results with hyperspectral images collected by the National Aeronautics and Space Administration Jet Propulsion Laboratory's Airborne Visible/Infrared Imaging Spectrometer and the Reflective Optics Spectrographic Imaging System indicate that the proposed scheme provides better performance when compared with other widely used methods.

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 letter, an ensemble learning approach, Rotation Forest, has been applied to hyperspectral remote sensing image classification for the first time. The framework of Rotation Forest is to project the original data into a new feature space using transformation methods for each base classifier (decision tree), then the base classifier can train in different new spaces for the purpose of encouraging both individual accuracy and diversity within the ensemble simultaneously. Principal component analysis (PCA), maximum noise fraction, independent component analysis, and local Fisher discriminant analysis are introduced as feature transformation algorithms in the original Rotation Forest. The performance of Rotation Forest was evaluated based on several criteria: different data sets, sensitivity to the number of training samples, ensemble size and the number of features in a subset. Experimental results revealed that Rotation Forest, especially with PCA transformation, could produce more accurate results than bagging, AdaBoost, and Random Forest. They indicate that Rotation Forests are promising approaches for generating classifier ensemble of hyperspectral remote sensing.

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.

Endmember extraction is an important task for hyperspectral analysis; the accurate identification of endmembers enables efficient spectral unmixing and classification. In the paper, a new endmember extraction algorithm based on a modified MSS approach with LP error as initial spectrum selection algorithm, and OPD measure as similarity is proposed. The endmembers extracted by modified MSS are more similar than that of MSS algorithm; from the experiments results, it has proved that our proposed method outperforms the existed MSS and N-FINDR algorithms.

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 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.

Support vector machine (SVM) and Random Forest (RF) have been developed to improve the accuracy of hyperspectral remote sensing (HRS) image classification significantly in recent years. Due to the different characteristics and obvious diversity between SVM and RF, we propose two integration approaches which combine SVM and Random Forest to classify the HRS image. The proposed method called DWDCS is examined by two hyperspectral images and it can acquire the higher overall accuracy and also improve the accuracy of each classes. Experimental results indicate that the proposed approaches have a great deal of advantages in classifying HRS image.

In order to compare fusion algorithms considering both active and passive remotely sensed data, a few well-known techniques, including Brovey, Gram-Schmidt spectral sharpening (GS), Hue Saturation Value (HSV), Principal Component Analysis spectral sharpening (PCA), and à trous wavelet transform applied in the Hue Intensity Saturation space (ATWT+HIS) are compared with the simple joint analysis of the original SAR and optical images. Experiments are performed using pairs of ALOS ANVIR-2 and PALSAR, SPOT and PALSAR, Landsat TM and ERS data. In this paper, the above mentioned methods and dataset combinations are tested and compared by means of quantitative indexes such as entropy, average gradient (AG), correlation coefficient (CC), deviation index (DI) and classification accuracy. The results obtained demonstrate that classification accuracy values can be improved by using these approaches by as much as 10% with respect to the best achievable value using only optical and SAR data separately. By means of a detailed analysis of the relationship between classification accuracy and quantitative indexes usually considered to evaluate the value of the fused products, our experiments show that larger deviation index (DI) and smaller correlation coefficient (CC) values are usually connected to more accurate classification results.

In this paper, ASTER images are processed from two aspects: the thermal infrared data are used to estimate land surface temperature (LST), and the multi-spectral data are used for gaining land cover information. The relationship between LST and corresponding land cover is obtained. GIS is also used to simulate the evolution of thermal environment and predict its trends under the specific land cover scenarios. The results demonstrate that the simulation result is credible and it is helpful to analyze land use planning schemes from specific aspects or urban thermal environment view, so land use planning could be improved by optimizing thermal environment and avoiding the negative impacts resulted from urban heat island.

Decision level fusion, using a specific criterion or algorithm to integrate the classified results from different classifiers, has shown great benefits to improve classification accuracy of multi-source remote sensing images. In this paper, three decision level fusion methods and four schemes for input data are used to hyperspectral remote sensing image classification. Different feature combination and decision level fusion approaches are experimented and analyzed, and the results show that decision level fusion is effective to improve the performance of hyperspectral remote sensing image classification.

A framework of multi-objective processing and analysis of ASTER data for urban environmental analysis was proposed at first, and related key techniques were discussed. Five levels of multi-objective information processing were discussed. The first level is visualization-level combination, in which the simple combination of results of different objectives is used for visualization or better representation. The second level is coefficient-level combination, and it aims to derive some new parameters by the combination of results for different objectives. The third level is collaboration-level combination, in which the output from one process is feed into the other processes and to improve the performance. The fourth level is association-level combination with the aim of finding those association rules among the results of different processes by statistics and data mining. The fifth level is collaboration decision-level integration, and this level aims to monitor, analyze, manage and protect urban environment by integrating above data, rules, parameter and models. The multi-objective processing and analysis of ASTER data will bring more benefits for urban environment Remote Sensing applications than ever, and this framework and related information processing methods could be used to other fields further.

Normalized difference vegetation index (NDVI), as the most important index derived from the red and near infrared spectrum scope of multi-spectral remotely sensed data, plays important roles to remote sensing applications in different fields. Three data sources operated by different vendors are chosen to compare the NDVIs derived from them and discover mutual relationships. The three sensors are ASTER, CBERS and Landsat ETM+. Before NDVIs are computed, image registration, geometric correction and atmospheric correction are conducted. The comparison is conducted from three levels: global level at the whole image scene, local level at some specific areas, and special object level by using related statistical indexes, and some useful suggestions are given based on the comparison.

Based on the analysis to DM and hyperspectral RS information processing, it is pointed out that hyperspectral RS data mining will promote the development of intelligent information processing. The framework and some key techniques are discussed in detail. The knowledge that can be discovered from hyperspectral RS information includes: spectral signatures of ground objects; spatial characteristics, rules and relationships; knowledge about genesis, characterization and diagnosis; relationship among different bands and spectral knowledge; dynamic evolution knowledge and abnormity and isolated point identification. By analysis and experiments, some algorithms proved effective to HRSDM including association rule mining, clustering and artificial neural network (ANN), rough set and fuzzy theory, decision tree and data cube are analyzed. Finally, some potential applications of HRSDM including typical information extraction and identification, quantitative RS and RS inversion, image classification and mixed pixel decomposition, and feature extraction and selection of optimal band combination are discussed.

Some basic issues on content-based remote sensing image retrieval are discussed in this paper. The framework, processing flow and levels are proposed based on theory of CBIR and characteristics of RS image. Oriented to the practical demands, five retrieval patterns including template-based, attribute-based, metadata-based, semanteme-based and integrated retrieval are proposed. The contents and features that can be used in content-based remote sensing image retrieval include color, shape, texture, spectra, spatial relation, metadata and relative rules and knowledge. Among those features, spectral features, spatial features and metadata are the main aspects of RS image differing from common images.