Pansharpening of multispectral images aims to merge multispectral (MS) and panchromatic (PAN) images to produce high-resolution multispectral (HRMS) images. Unsupervised pansharpening algorithms, which are widely used in deep learning for pansharpening tasks, commonly employ generative adversarial networks (GANs). However, existing unsupervised methods based on GANs have some limitations: 1) restricted ability for joint preservation of spatial and spectral information, and 2) the receptive field of general convolutions restricts the extraction of long-range dependencies. To address these issues, we propose a frequency-guided generative adversarial networks for unsupervised pansharpening (FG-GAN). Our unsupervised framework uses high-frequency and low-frequency information as prior constraints to guide the training of the FG-GAN’s generator and discriminator networks, thereby enhancing the joint preservation of spatial and spectral details. Furthermore, a graph convolution-based generator network is designed, in which long-range edge dependencies are extracted and propagated by learning the relationships between distant edge feature nodes. Extensive experiments on the Quickbird and Gaofen-2 datasets demonstrate the effectiveness of our design: our method enhances spatial details and reduces the spatial distortion index (Ds) by 44.8%, while achieving the fidelity of spatial-spectral information with a HQNR score of 0.99.

The joint clustering of multimodal remote sensing (RS) data represents a multiobjective optimization challenge involving conflicting modality-specific objectives and diverse regularization objectives. Current approaches to multiview subspace clustering (MVSC) often oversimplify this task by transforming it into a weighted single-objective optimization problem, neglecting the intricate interactions between objectives and leading to suboptimal subspace representations. The presence of quadratic decision variables in MVSC renders direct application on large-scale RS data impracticable using multiobjective evolutionary algorithms (MOEAs). To overcome this challenge, we propose a novel MVSC method termed multiobjective multiview attributed graph learning (MMAGL). Instead of optimizing every self-representation coefficient individually, our method transforms MVSC into a link prediction task over a sparse attributed graph that fuses different modalities. We incorporate superpixel-based sample reduction and proximity-based population coding, leveraging spatial and structural priors, respectively. This results in a significantly compressed decision space, enabling optimization with MOEAs. To fully exploit node attributes and the graph structure, we redefine self-representation using contrastive learning and introduce an efficient graph filtering (GF) through a generalized spectral graph convolution, enhancing clustering discriminability. The proposed MMAGL constitutes a hybrid and versatile framework, adaptable to any MOEA. Extensive experimental evaluations demonstrate that our MMAGL method surpasses the current state-of-the-art on multimodal RS benchmarks (e.g., with nearly 2% gain on Trento and 3% on Houston) on overall accuracy.

Self-supervised hyperspectral image (HSI) clustering remains a fundamental yet challenging task due to the absence of labeled data and the inherent complexity of spatial-spectral interactions. While recent advancements have explored innovative approaches, existing methods face critical limitations in clustering accuracy, feature discriminability, computational efficiency, and robustness to noise, hindering their practical deployment. In this paper, a self-supervised efficient low-pass contrastive graph clustering (SLCGC) is introduced for HSIs. Our approach begins with homogeneous region generation, which aggregates pixels into spectrally consistent regions to preserve local spatial-spectral coherence while drastically reducing graph complexity. We then construct a structural graph using an adjacency matrix A and introduce a low-pass graph denoising mechanism to suppress high-frequency noise in the graph topology, ensuring stable feature propagation. A dual-branch graph contrastive learning module is developed, where Gaussian noise perturbations generate augmented views through two multilayer perceptrons (MLPs), and a cross-view contrastive loss enforces structural consistency between views to learn noise-invariant representations. Finally, latent embeddings optimized by this process are clustered via K-means. Extensive experiments and repeated comparative analysis have verified that our SLCGC contains high clustering accuracy, low computational complexity, and strong robustness. The code source will be available athttps://github.com/DY-HYX.

The joint clustering of multimodal remote sensing (RS) data poses a critical and challenging task in Earth observation. Although recent advances in multiview subspace clustering have shown remarkable success, existing methods become computationally prohibitive when dealing with large-scale RS datasets. Moreover, they neglect intrinsic nonlinear and spatial interdependencies among heterogeneous RS data and lack generalization ability for out-of-sample data, thereby restricting their applicability. This article introduces a novel unified framework called anchor-based multiview kernel subspace clustering with spatial regularization (AMKSC). It learns a scalable anchor graph in the kernel space, leveraging contributions from each modality instead of seeking a consensus full graph in the feature space. To ensure spatial consistency, we incorporate a spatial smoothing operation into the formulation. The method is efficiently solved using an alternating optimization strategy, and we provide theoretical evidence of its scalability with linear computational complexity. Furthermore, an out-of-sample extension of AMKSC based on multiview collaborative representation-based classification is introduced, enabling the handling of larger datasets and unseen instances. Extensive experiments on three real heterogeneous RS datasets confirm the superiority of our proposed approach over state-of-the-art methods in terms of clustering performance and time efficiency. The source code is available at https://github.com/AngryCai/AMKSC.

Transformer-based methods have a great ability to model nonlocal interactions between spectral and spatial information, while the local features are easily ignored. Graph convolutional neural networks (GCNs) tend to do well in exploiting neighborhood vertex interactions based on their unique aggregation mechanism, while the ability to extract global information is limited. In this article, we study to comprehensively utilize the advantages of transformer and graph convolution by combining the two structures into a unified Transformer (Graphormer) to construct both local and global interactions for hyperspectral image (HSI) classification, and spatial–spectral features enhanced Graphormer framework (S2GFormer) is proposed. Specifically, a follow patch mechanism is first proposed to transform the pixel in HSI to patches while preserving the local spatial features and reducing the computational cost. Moreover, a patchwise spectral embedding block is designed to extract the spectral features of the patch, in which a neighborhood convolution is inserted for comprehensive spectral information extraction. Finally, a multilayer Graphormer Encoder module is proposed to extract the representative spatial–spectral features from the patch for HSI classification. In our network, we jointly integrate the three aforementioned parts into a unified network, and each component benefits the other. The experimental results demonstrate its suitability for HSI classification when compared with other state-of-the-art (SOTA) classifiers, particularly in scenarios with very limited labeled samples. The code of S2GFormer will be made publicly available at:https://github.com/DY-HYX.

Efficient extraction of spectral sequences and geospatial information is crucial in hyperspectral image (HSI) classification. Recurrent neural networks (RNNs) and Transformers excel in capturing long-range spectral features, while convolutional neural networks (CNNs) excel in aggregating spatial information through convolutional kernels. However, RNNs and Transformers suffer from low-computational efficiency, and CNNs have limitations in perceiving global contextual information. To address these issues, this article proposes GraphMamba—an efficient graph structure learning vision Mamba for HSI classification. Specifically, GraphMamba is a novel hyperspectral information processing paradigm that preserves spatial-spectral features by constructing spatial-spectral cubes and employs a linear spectral encoder to enhance the operability of subsequent tasks. The core components of GraphMamba include the HyperMamba module, which enhances computational efficiency, and the SpatialGCN module, designed for adaptive spatial context awareness. The HyperMamba mitigates clutter interference by employing a global mask (GM) and introduces a parallel training and inference architecture to alleviate computational bottlenecks. Meanwhile, the SpatialGCN utilizes weighted multihop aggregation (WMA) for spatial encoding, emphasizing highly correlated spatial structural features. This approach enables flexible aggregation of contextual information while minimizing spatial noise interference. Notably, the encoding modules of the proposed GraphMamba architecture are both flexible and scalable, providing a novel approach for the joint mining of spatial-spectral information in hyperspectral images. Extensive experiments were conducted on three different scales of real HSI datasets. When compared with state-of-the-art classification methods, GraphMamba demonstrated superior performance. The core code will be released athttps://github.com/ahappyyang/GraphMamba.

This article presents FLGC, a simple yet effective fully linear graph convolutional network for semi-supervised and unsupervised learning. Instead of using gradient descent, we train FLGC based on computing a global optimal closed-form solution with a decoupled procedure, resulting in a generalized linear framework and making it easier to implement, train, and apply. We show that (1) FLGC is powerful to deal with both graph-structured data and regular data, (2) training graph convolutional models with closed-form solutions improve computational efficiency without degrading performance, and (3) FLGC acts as a natural generalization of classic linear models in the non-Euclidean domain (e.g., ridge regression and subspace clustering). Furthermore, we implement a semi-supervised FLGC and an unsupervised FLGC by introducing an initial residual strategy, enabling FLGC to aggregate long-range neighborhoods and alleviate over-smoothing. We compare our semi-supervised and unsupervised FLGCs against many state-of-the-art methods on a variety of classification and clustering benchmarks, demonstrating that the proposed FLGC models consistently outperform previous methods in terms of accuracy, robustness, and learning efficiency. The core code of our FLGC is released at https://github.com/AngryCai/FLGC .

Deep neural networks have gained increasing interest in hyperspectral image (HSI) processing. However, prior arts often neglect the high-order correlation among data points, failing to capture intraclass variations. In this letter, we present a unified neural network framework, termed as hypergraph-structured autoencoder (HyperAE), to leverage the high-order relationship among data and learn robust deep representation for downstream tasks. Technically, the proposed method adopts a deep autoencoder regularized by hypergraph structure as the backbone network, which is jointly trained with a task-specific branch, resulting in a multitask architecture. We separately combine the subspace clustering model and the softmax classifier into the HyperAE to deal with HSI unsupervised and semisupervised classification problems. Benefiting from the hypergraph, HyperAE endows traditional networks with the capacity of preserving the high-order structured information. We evaluate the proposed methods on three benchmarking HSI data sets, demonstrating that the proposed HyperAE dramatically outperforms many existing methods with significant margins in both unsupervised and semisupervised HSI classification problems.

Hyperspectral image (HSI) clustering has attracted a great deal of attention, owing to lower cost and higher application prospects. Deep subspace clustering has been proved to be an effective method to explore the sample relationship of HSI clustering. However, due to the complex distribution of HSI data, the problem of data cluster overlap occurs frequently. In the actual sample distribution, a sample may belong to multiple subspaces. The complex sample distribution brings challenges to subspace clustering. In this letter, we propose a deep mutual information subspace clustering network (DMISC) to find a more intuitive feature space for non-linear subspace clustering. Technically, we maximize the mutual information between the samples and their generated features to enlarge the inter-class dispersion and intra-class compactness. The deep subspace method can find a more suitable non-linear intrinsic relationship, benefitting from the generated feature distribution. We evaluate DMISC on four HSI data sets and compare the performances with 12 popular clustering methods. The experiment results demonstrate our method outperforms many prior unsupervised methods.

Deep subspace clustering has achieved remarkable performances in the unsupervised classification of hyperspectral images. However, previous models based on pixel-level self-expressiveness of data suffer from the exponential growth of computational complexity and access memory requirements with increasing number of samples, thus leading to poor applicability to large hyperspectral images. This paper presents a Neighborhood Contrastive Subspace Clustering network (NCSC), a scalable and robust deep subspace clustering approach, for unsupervised classification of large hyperspectral images. Instead of using a conventional autoencoder, we devise a novel superpixel pooling autoencoder to learn the superpixel-level latent representation and subspace, allowing a contracted self-expressive layer. To encourage a robust subspace representation, we propose a novel neighborhood contrastive regularization to maximize the agreement between positive samples in subspace. We jointly train the resulting model in an end-to-end fashion by optimizing an adaptively weighted multi-task loss. Extensive experiments on three hyperspectral benchmarks demonstrate the effectiveness of the proposed approach and its substantial advancement of state-of-the-art approaches.

Due to prior knowledge deficiency, large spectral variability, and high dimension of hyperspectral image (HSI), HSI clustering is extremally a fundamental but challenging task. Deep clustering methods have achieved remarkable success and have attracted increasing attention in unsupervised HSI classification (HSIC). However, the poor robustness, adaptability, and feature presentation limit their practical applications to complex large-scale HSI datasets. Thus, this article introduces a novel self-supervised locality preserving low-pass graph convolutional embedding method (L2GCC) for large-scale hyperspectral image clustering. Specifically, a spectral–spatial transformation HSI preprocessing mechanism is introduced to learn superpixel-level spectral–spatial features from HSI and reduce the number of graph nodes for subsequent network processing. In addition, locality preserving low-pass graph convolutional embedding autoencoder is proposed, in which the low-pass graph convolution and layerwise graph attention are designed to extract the smoother features and preserve layerwise locality features, respectively. Finally, we develop a self-training strategy, in which a self-training clustering objective employs soft labels to supervise the clustering process and obtain appropriate hidden representations for node clustering. L2GCC is an end-to-end training network, which is jointly optimized by graph reconstruction loss and self-training clustering loss. On Indian Pines, Salinas, and University of Houston 2013 datasets, the clustering accuracy overall accuracies (OAs) of the proposed L2GCC are 73.51%, 83.15%, and 64.12%, respectively.

Hyperspectral anomaly detection (HAD) aims to distinguish anomalies from background-by-background modeling. Deep learning has been applied to HAD and achieves promising detection results. However, there exist several issues that need to be addressed: 1) unrealistic Gaussian assumption on the latent representations may limit its application; 2) deep features are not well-suited to anomaly detection due to the separation between feature learning and anomaly detection; 3) lack of adequate exploitation of spectral-spatial features; 4) negative effect caused by spectral band redundancy. In this article, we propose an end-to-end trainable deep one-class classification network for HAD. Specifically, a minimal enclosing hypersphere is trained to involve the deep features of background samples. These background samples are selected by a density clustering-based method. In this way, feature learning and anomaly detection are incorporated into a unified framework. Meanwhile, there is no explicit Gaussian assumption on the background features. Moreover, due to the complementarity of spectral and spatial features, a novel feature fusion strategy is proposed to fuse spectral and spatial features extracted by a two-stream deep convolutional autoencoder network. Finally, a band attention module is used to automatically learn small weights for redundant bands and thus reduce the negative effect caused by redundant bands. Experimental results on five public datasets demonstrate the superiority of the proposed method compared to several state-of-the-art HAD methods in the detection performance.

Randomized neural networks (NNs), such as random vector functional link (RVFL) and extreme learning machine (ELM), have been widely applied in various classification problems owing to their computational efficiency and universal approximation capability. However, such approaches are designed for regular Euclidean data and lack the ability to generalize to complex structured data. Moreover, their randomly generated parameters often lead to a suboptimal decision boundary with a growing requirement of hidden neurons. In this article, we first propose a plain framework, termed randomized graph convolutional networks (RGCNs), to generalize the classic randomized NNs to the non-Euclidean domain. Then, a hybrid framework called evolution-driven RGCN (EvoRGCN) is presented by using adaptive differential evolution with novelty search strategy to seek the globally optimal graph embedding for the plain RGCN. Finally, we recast the classic ELM and RVFL under the proposed frameworks, resulting in four novel semi-supervised models, including the plain models [i.e., graph convolutional extreme learning machines (GCELMs) and graph convolutional RVFL (GCRVFL)] and the optimized models (i.e., O-GCELM and O-GCRVFL). We show that our approaches are the natural generalization of the traditional randomized NNs in the non-Euclidean domain. Furthermore, our approaches not only retain the advantages of the classic approaches but also enable them to handle graph data. We compare our approaches against many existing methods across regular datasets and graph benchmarks, demonstrating that the proposed approaches dramatically outperform the compared methods with better generalization ability and robustness. Particularly, we quantitatively show the performance ranking of different randomized NNs, i.e., O-GCRVFL$> $O-GCELM$\approx $GCRVFL$> $GCELM$\approx $RVFL$> $ELM.

Recently, deep learning-based methods have made great progress in hyperspectral image (HSI) classification (HSIC). Different from ordinary images, the intrinsic complexity of HSIs data still limits the performance of many common convolutional neural network (CNN) models. Thus, the network architecture becomes more and more complex to extract discriminative spectral-spatial features. For instance, 3-D CNN usually has a large number of trainable parameters, thus increasing the computational complexity of the HSIC. In this letter, we designed a cooperative spectral-spatial attention dense network (CS2ADN) that takes raw 3-D HSI data as input data. Specifically, the attention module consists of spectral and spatial axes, by which the salient spectral-spatial features will be emphasized. Furthermore, we combined these attention modules with the dense connection, which is termed as the lightweight dense block; it has a lower computation cost and achieves better classification performance. At the same time, we introduced the center loss, by jointly using the supervision of the center loss and the softmax loss, where the discriminative features could be clearly observed, particularly for small data sets. Experimental results on the biased and unbiased HSI data show that our method outperforms several state-of-the-art methods in HSIC with small training samples.

Hyperspectral image (HSI) clustering is a challenging task due to the high complexity of HSI data. Subspace clustering has been proven to be powerful for exploiting the intrinsic relationship between data points. Despite the impressive performance in the HSI clustering, traditional subspace clustering methods often ignore the inherent structural information among data. In this article, we revisit the subspace clustering with graph convolution and present a novel subspace clustering framework called graph convolutional subspace clustering (GCSC) for robust HSI clustering. Specifically, the framework recasts the self-expressiveness property of the data into the non-Euclidean domain, which results in a more robust graph embedding dictionary. We show that traditional subspace clustering models are the special forms of our framework with the Euclidean data. On the basis of the framework, we further propose two novel subspace clustering models by using the Frobenius norm, namely efficient GCSC (EGCSC) and efficient kernel GCSC (EKGCSC). Each model has a globally optimal closed-form solution, making it easier to implement, train, and apply in practice. Extensive experiments strongly evidence that EGCSC and EKGCSC dramatically outperform current models on three popular HSI data sets consistently.

Extreme Learning Machine (ELM) has gained lots of research interest due to its universal approximation capability and fast learning speed. However, traditional ELMs are devised for regular Euclidean data, such as 2D grid and 1D sequence, and thus don't apply to non-Euclidean data, e.g., graph-structured data. To overcome this shortcoming, this paper presents a Graph Convolutional Extreme Learning Machine (termed as GCELM) for semi-supervised classification. Technically, a random graph convolutional layer is introduced to replace the random projection of original ELM, which endues ELM with the capability of dealing with graph-structured data directly. To generate a robust graph from the raw dataset, a self-representation model is adopted to construct a weighted graph. Extensive experiments on 27 UCI datasets demonstrate that GCELM outperforms many popular semi-supervised methods, and with faster learning speed. To the best of our knowledge, this is the first work that combines graph convolution with ELM.

Classification of hyperspectral images (HSIs) by making full use of the spectral and the spatial information has become a research hotspot in the field of remote sensing technology. Aiming at the problems of information redundancy and low utilization of spatial information, this letter proposes a visual saliency-based extended morphological profile (VS-EMP) scheme. First, the morphological features are extracted by the EMP from the HSIs on several principal components. Second, the local binary pattern (LBP) is performed to extract the texture features from morphological scenes. Third, saliency features are captured according to the texture features in an approach of Boolean mapping saliency (BMS). Finally, spectral-spatial features are constructed by feature fusion and are further used for the classification of the HSIs. A number of experiments are performed, including using different classifiers to verify the performance of the proposed scheme, comparing with related variant algorithms, comparing time with deep learning, and testing learning ability in the absence of labeled samples. Experimental results indicate that the proposed method is significantly superior to the previous methods.

Hyperspectral image (HSI) consists of hundreds of continuous narrowbands with high spectral correlation, which would lead to the so-called Hughes phenomenon and the high computational cost in processing. Band selection (BS) has been proven to be effective in avoiding such problems by removing redundant bands. However, many existing BS methods separately estimate the significance for every single band and cannot fully consider the nonlinear and global interaction between spectral bands. In this article, by assuming that a complete HSI band set can be reconstructed from its few informative bands, we propose a unified BS framework, BS Network (BS-Net). The framework consists of a band attention module (BAM), which aims to explicitly model the nonlinear interdependences between spectral bands, and a reconstruction network (RecNet), which is used to restore the original HSI from the learned informative bands, resulting in a flexible architecture. The resulting framework is end-to-end trainable, making it easier to train from scratch and to combine with many existing networks. We implement two versions of BS-Nets, respectively, using fully connected networks (BS-Net-FC) and convolutional neural networks (BS-Net-Conv), and extensively compare their results with popular existing BS approaches on three real hyperspectral data sets, showing that the proposed BS-Nets can accurately select informative band subset with less redundancy and outperform the competitors in terms of classification accuracy with competitive time cost.

This paper presents a novel clustering method, named Laplacian regularized deep subspace clustering (LRDSC), for unsupervised hyperspectral image (HSI) classification. We introduce the Laplacian regularization into the subspace clustering to consider the manifold structure reflecting geometric information. To enable the subspace clustering, which works in linear space, to deal with the complicated HSI data with nonlinear characteristics, we combine the subspace clustering as a self-expressive layer with deep convolutional auto-encoder. Furthermore, the 3-D convolutions and deconvolutions with skip connections are utilized to make full extraction of the spectral-spatial information and full use of the historical feature maps produced by the network. We compare the results of the proposed method with six existing cluster methods on four real hyperspectral data sets, showing that the proposed method is able to achieve state-of-the-art performance.

Hyperspectral images (HSIs) consist of hundreds of continuous bands with high correlation, making it contain great abundant information. Band selection is an effective idea for removing redundant bands and preserving the physical significance at the same time. Popular sparse representation-based band selection commonly introduces an additional coefficient to combine error term and sparse constraint term, making it difficult to find out the optimal balance coefficient. In this letter, we propose a hybrid clustering-based band-selection approach based on using evolutionary multiobjective optimization to solve a sparse self-representation model constituted with two conflicting objectives. The proposed approach simultaneously minimizes two terms of the sparse representation model, avoiding the balance coefficient and producing a set of optimal solutions that are used to construct a similarity matrix for spectral clustering. Finally, a reduced band subset is determined by the cluster centers. We compare the results of the proposed approach with four existing band-selection methods for three real HSI data sets, showing that the proposed approach is able to effectively select representative bands with better classification accuracy.

Hyperspectral image (HSI) consists of hundreds of continuous narrow bands with high redundancy, resulting in the curse of dimensionality and an increased computation complexity in HSI classification. Many clustering-based band selection approaches have been proposed to deal with such a problem. However, a few of them consider the spectral and spatial relationship simultaneously. In this letter, we proposed a novel clustering-based band selection approach using deep subspace clustering (DSC). The proposed approach combines the subspace clustering task into a convolutional autoencoder by treating it as a self-expressive layer, enabling it to be trained end to end. The resulting network can fully extract the interaction of spectral bands based on using spatial information and nonlinear feature transformation. We compared the results of the proposed method with existing band selection methods for three widely used HSI data sets, showing that the proposed method is able to accurately select an informative band subset with remarkable classification accuracy.

Currently, deep neural networks (DNNs) are an important method for handling hyperspectral image (HSI) classification because of their good performance in image processing. However, DNNs' performance depends on a massive number of training data and hyperparameters that are carefully fine-tuned, which results in structural complexity and a time-consuming process. Deep forest is a novel deep learning method that does not need much training data and has a simple structure. In this paper, we first design a deep forest for spectral-based HSI classification and then propose an improved deep forest algorithm, named deep multigrained cascade forest (dgcForest), for spatial-based HSI classification. On the one hand, the cascade forest in dgcForest is used in multigrained scanning, which enhances the performance; on the other hand, a pooling layer is added after the multigrained scanning to reduce the output dimensions. To demonstrate that our proposed algorithm presents a good performance in HSI classification, we analyze the hyperparameters of deep forest and dgcForest and compare them with other methods on the biased and unbiased data sets, which illustrates that our method is superior to other state-of-the-art deep learning methods.

Deep neural networks have been successfully applied to many data mining problems in recent works. The training of deep neural networks relies heavily upon gradient descent methods, however, which may lead to the failure of training due to the vanishing gradient (or exploding gradient) and local optima problems. In this paper, we present SEvoAE method based on using Evolutionary Multiobjective optimization (EMO) algorithm to train single layer auto-encoder, and sequentially learning deeper representation in a stacking way. SEvoAE is able to achieve accurate feature representation with good sparseness by globally simultaneously optimizing two conflicting objective functions and allows users to flexibly design objective functions and evolutionary optimizers. We compare results of the proposed method with existing architectures for seven classification problems, showing that the proposed method is able to outperform existing methods with a reduced risk of overfitting the training data.