Tiny object detection (TOD) in remote sensing imagery remains challenging because foreground signals are extremely weak in deep feature hierarchies and are easily overwhelmed by high-response background interference. To mitigate this observed foreground-background signal modulation imbalance (FBSMI) difficulty, we propose a signal modulation network (SMN) for remote-sensing TOD. SMN comprises two complementary components. First, an adaptive Wiener filter modulator (AWFM) is inserted after backbone stages to suppress background-dominated noise while preserving weak target-related responses at multiple resolutions. Second, we introduce the novel denoising diffusion transformer (DDT), a feature-space conditional diffusion module that operates on detector feature tensors rather than image pixels. DDT generates multiple diffusion-guided semantic feature variants from high-level fused features and expands the local representation space around weak tiny object evidence. Extensive experiments on AI-TOD, SODA-A, DOTAv2.0, and DIOR-R demonstrate that SMN not only effectively mitigates the FBSMI problem, but also improves detection accuracy, particularly for very tiny and tiny objects, compared with state-of-the-art methods.

Efficient tiny object detection (TOD) in large-size remote sensing imagery (LSRSI) is particularly challenging in real-world remote sensing applications. We observe that as the input size of the remote sensing scene increases, TOD faces more severe foreground signal identification issues. To address this, we are the first to design a backbone network from the perspective of low-level spatial feature preservation and utilization, specifically for tiny object feature extraction in large-size remote sensing scene patches. The proposed architecture, referred to as the resolution preserving and utilization network (RPUN), demonstrates excellent foreground tiny object feature response identification ability when increasing the input size of remote sensing scenes, effectively maintaining detection performance comparable to that of smaller input slices. Additionally, we introduce GF2UBSv2, a large-scale panchromatic satellite imagery dataset focused on tiny urban bridge detection. Extensive experiments conducted on GF2UBSv2, DIOR, SODA-A, and DOTAv2.0 demonstrate the superior performance of RPUN compared with state-of-the-art methods. The code and dataset are available at: https://github.com//Nankle.

The advancement of spaceborne hyperspectral remote sensing technology has led to the widespread use of hyperspectral imaging, due to its ability to detect subtle spectral differences. Most of the traditional machine learning (ML) methods and popular deep learning (DL) architectures for hyperspectral image (HSI) classification either fail to capture global features or demand high computational resources. While multilayer perceptron (MLP)-based models offer a computationally efficient alternative, they struggle to capture manifold structures and are susceptible to overfitting. To address these challenges, we propose a novel spatial-spectral hypergraph dynamic gating MLP (S2H-DGMLP) framework tailored for HSI classification. The spatial–spectral hypergraph enhances discriminative power by modeling high-order spatial and spectral correlations, jointly optimizing local spatial features and global spectral features to produce more separable feature representations in the embedding space. Within this framework, the channel and spatial projections are statically parameterized using MLP, while the dynamic gating MLP (DGMLP) block captures global contextual information. The dynamic gating mechanism within the DGMLP block automatically adjusts the segmentation ratio to balance spatial and spectral contributions, while incorporating complex nonlinear combinations to improve feature representation. Experimental results on the Pavia University and Houston datasets demonstrate that S2H-DGMLP significantly improves classification performance, confirming its effectiveness in HSI classification tasks.

Since the rapid development of deep learning, the task of cloud removal in multispectral optical (MSO) images has been revisited and reconsidered from new perspectives. However, due to insufficient attention from researchers, the study of Synthetic Aperture Radar (SAR)-based cloud removal methods based on deep learning remains in its early stages. Crude image-level fusion strategies are often adapted for SAR-based MSO image cloud removal. This study proposes a novel Transformer-based cloud removal network (TCRNet) utilizing SAR edge features to enhance the recovered region in MSO images. TCRNet models each spectral channel as a token, enabling efficient self-attention computations across the spectral dimension. This strategy empowers the proposed model to exploit the global spectral features captured by MSO sensors. A SAR edge feature extraction module is designed to extract multi-scale edge prior maps. The resulting features are seamlessly integrated into the U-shape MSO backbone network using spectral-wise multi-head self-attention, elevating attention to the critical edge regions during feature fusion. During the training process,Lmaeloss constrains the pixel-level approximation, andLperspatial perceptual loss enhances the similarity of texture features between the predicted cloud-free and ground truth MSO images, regardless of cloud coverage. Experiments conducted on the SEN12MS-CR Dataset show that TCRNet outperforms state-of-the-art methods. The source code will be released at https://github.com/MinghuaWang123/TCRNet.

Linear regression and its variants have achieved considerable success in image classification. However, those methods still encounter two challenges when dealing with hyperspectral image (HSI) classification. On the one hand, the existing ones focus on mining the relationship between the label space and original data space during the classifier training, which is generally sensitive to noise corruptions. On the other hand, transforming the training samples into a strict binary label matrix makes the generalization ability of the classifier limited. To address these challenges, this paper constructs a novel integrative model called tensor decomposition-based relaxed linear regression (TDRLR) for HSI classification. Firstly, the model adopts tensor canonical polyadic (CP) decomposition to learn two dictionaries from spatial and spectral directions respectively, which can help to generate a double dictionary representation for HSI data. Then, the linear regression classifier is integrated to learn a transformation that reveals the mapping relation between the double dictionary representation and label space rather than the original data for enhancing robustness. Meanwhile, a more flexible way, label relaxation, is employed to enlarge the margins between different classes. More importantly, the learned double dictionary representation and classifier can be fine-tuned in tandem to enhance performance through the designed alternate iterative jointly learning algorithm. Experiments conducted on four real-world HSI datasets demonstrate that the proposed method achieves significant improvements in classification performance with a small size training set, when compared with state-of-the-art HSI classification methods.

Hyperspectral anomaly detection (HAD) seeks to identify targets of interest within hyperspectral images (HSIs) without requiring prior knowledge. Recent works show that embedding low-rank attributes in the HAD task can yield superior results. While most such methods focus on exploiting global or local geometrical structures in HSIs, largely overlooking similar features in nonlocal regions. To this end, we propose a novel nonlocal framework that captures the nonlocal self-similar (NSS) and global correlation along spectrum (GCS) priors within the subspace of HSIs. Then, we integrate a deep denoiser prior into the nonlocal framework to further leverage the external priors. The proposed approach, referred to as nonlocal and deep priors for anomaly detection (NLDPAD), combines the structural modeling of nonlocal low-rank and the feature extraction capabilities of deep learning (DL). Furthermore, an efficient alternating minimization (AM) algorithm was developed to optimize the framework, while the plug-and-play alternating direction method of multipliers (PnP-ADMM) was used to solve the subproblems. Extensive experiments on real satellite and aerial hyperspectral datasets demonstrate that the proposed NLDPAD framework significantly outperforms state-of-the-art methods, achieving superior detection performance.

Hyperspectral super-resolution (SR) can effectively enhance the spatial resolution of hyperspectral images, holding significant application value. Nevertheless, existing methods tend to overlook global information and some detailed aspects of hyperspectral images, resulting in limitations in feature extraction. In addressing this issue, we propose a Dilated Attention With Multi-Stream Learning (DAMS) network to facilitate the fusion of hyperspectral and multispectral images. The network comprises three autoencoders, incorporating dilated residual multipath feature extraction for high-resolution multispectral images and a dense convolutional neural network for low-resolution hyperspectral images. Notably, no prior knowledge of point spread function (PSF) and spectral response function (SRF) is required. Experimental results with DAMS showcase its advantages over other super-resolution fusion methods, demonstrating robust performance across diverse datasets with varying PSF and SRF.

Due to its powerful feature extraction and representation capabilities, deep learning has been successfully applied in the field of hyperspectral image classification. In patchbased hyperspectral image classification, where the central pixel represents the true category, extracting spectral features similar to the central pixel spectrum is crucial. To this end, we develop a new framework, called the inverse Mahalanobis attention network (IMAN), to address the spectral similarity feature extraction problem. The proposed framework develops a self-attention mechanism module based on the Mahalanobis distance to better learn the correlation between feature vectors of pixels, thereby efficiently suppressing noise generated by different land cover categories within patches. The feature extraction capability is further enhanced by integrating a dual-stream network structure that separates spatial and spectral information in hyperspectral images. Experiments conducted on on real hyperspectral datasets demonstrate the effectiveness and superiority of the proposed method compared to several state-of-the-art hyperspectral image classification methods.

In recent years, deep learning has been proposed for hyperspectral images(HSIs) super-resolution, and many fusion models for HSI and multispectral images(MSIs) have been developed. However, these networks are constrained to the structure of convolutional neural networks(CNNs), and more attention needs to be paid to the disadvantage of the restricted receptive field of CNNs, such that some of the distal information needs to be included in the process of acquiring features. This approach involves capturing large-scale spatial features through multi-head spatial attention and spectral features of MSI and HSI through multi-head spectral attention. Subsequently, the features are processed by convolution kernels of different scales compactly. The effectiveness and competitiveness of MTSANet are evaluated by comparing it with some state-of-the-art (SOTA) methods.

Hyperspectral anomaly detection aims at distinguishing targets of interest from the background without prior knowledge. Although low-rank representation (LRR)-based methods have been broadly applied in anomaly detection tasks, how to approximate the penalties in LRR-based methods more precisely is still a problem that needs to be further investigated. To this end, this article designs a unified nonconvex framework called hyperspectral anomaly detection via generalized shrinkage mappings (HADGSMs) to better approximate the LRR-based methods. The core of the proposed framework is to design new nonconvex penalties to approximate the group sparsity,$l_{0}$gradient, and low-rankness penalties in the LRR-based anomaly detection models, which can be efficiently minimized by means of generalized shrinkage mappings (GSMs). Then, an efficient alternating direction method of multipliers (ADMM) is developed to handle the proposed model. Experiments conducted on several real hyperspectral datasets demonstrate the superiority and effectiveness of the proposed framework in enhancing detection performance with respect to state-of-the-art methods.

International audience

Hyperspectral unmixing aims at estimating pure spectral signatures and their proportions in each pixel. In practice, the atmospheric effects, intrinsic variation of the spectral signatures of the materials, illumination, and topographic changes cause what is known as spectral variability resulting in significant estimation errors being propagated throughout the unmixing task. To this end, we developed a new method, called the orthogonal subspace unmixing (OSU), to address spectral variability by utilizing the orthogonal subspace projection. The proposed OSU method jointly performs orthogonal subspace learning and the unmixing process to find a more suitable subspace for unmixing. The orthogonal subspace projection encourages the representation held in the subspace to be more distinct from each other to remove the complex spectral variability in the subspace. Furthermore, an alternating minimization (AM) was designed to solve the resulting optimization problem. An efficient and convergent symmetric Gauss–Seidel alternating direction method of multipliers (sGS-ADMM), essentially a special case of the semiproximal alternating direction method of multipliers (SPADMM), was developed to solve the subproblem. Experiments conducted on one synthetic data and two real data demonstrate the effectiveness and superiority of the proposed framework in mitigating the effects of spectral variability with respect to classical linear unmixing methods or variability accounting approaches.

Efforts to enhance the detection accuracy of hyperspectral (HS) anomaly detection (AD) have been significant, but the impact of noise resulting from HS data acquisition and transmission has not been well studied. Furthermore, the separation of denoising and subsequent interpretation makes it challenging to evaluate and control the influence of noise on the detection results. To this end, we proposed a joint anomaly detection and noise removal (ADNR) paradigm called DSR-ADNR, which develops a double subspace representation method to obtain both denoised and detection results simultaneously. DSR-ADNR uses a low-dimensional orthogonal basis to represent HS images and extract distinctive features for AD. The feature matrix is represented by a dictionary-based low-rank subspace that captures the complex nature of the low-dimensional features. In each iteration, DSR-ADNR utilizes the nonlocal self-similarity of the feature matrix to remove noise and improve intermediate detection performance. Meanwhile, the progressive LR representation of the background and anomalies for the feature matrix upgrades the explicit LR expression of nonlocal self-similar patches for better denoising. The well-designed linearized alternating direction method of multipliers with an adaptive penalty (LADMAP) is utilized to solve the proposed DSR-ADNR. Extensive experiments on simulated and real-world data sets demonstrate the effectiveness of DSR-ADNR in the HS AD task under different noise cases.

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.

With the remarkable development of spectral unmixing, the sparse-representation-based approaches have emerged as a promising alternative. The sparse-representation-based approaches aim at finding the optimal subset of a spectral library that can optimally model each pixel of a given hyperspectral image in a semisupervised fashion. The classic sparse unmixing models are solved by the prime alternating direction method of multipliers (pADMMs). However, the computation task of pADMM is heavy and time consuming. In this letter, we design a novel dual-alternating direction method of multipliers (dADMMs) for the classic sparse unmixing models. We also present the global convergence analysis of our algorithm in some special cases. As shown in our experiments, the proposed algorithm is more effective than the state-of-the-art algorithms.