Hyperspectral image classification (HSIC) is a valuable method for identifying coastal wetland vegetation, but challenges like environmental complexity and difficulty in distinguishing land cover types make large-scale labeling difficult. Cross-domain few-shot learning (CDFSL) offers a potential solution to limited labeling. Existing CDFSL HSIC methods have made significant progress, but still face challenges like prototype deviation, covariate shifts, and rely on complex domain alignment (DA) methods. To address these issues, a feature reconstruction-based CDFSL (FRFSL) algorithm is proposed. Within FRFSL, a Prototype Calibration Module (PCM) is designed for the prototype deviation, which employs a Bayesian inference-enhanced Gaussian Mixture Model to select reliable query features for prototype reconstruction, aligning the prototypes more closely with the actual distribution. Additionally, a ridge regression closed-form solution is incorporated into the Distance Metric Module (DMM), employing a projection matrix for prototype reconstruction to mitigate covariate shifts between the support and query sets. Features from both source and target domains are reconstructed into dynamic graphs, transforming DA into a graph matching problem guided by optimal transport theory. A novel shared transport matrix implementation algorithm is developed to achieve lightweight and interpretable alignment. Extensive experiments on three self-constructed coastal wetland datasets and one public dataset show that FRFSL outperforms eleven state-of-the-art algorithms. The code will be available at https://github.com/Yqx-ACE/TIP_2025_FRFSL.

Domain adaptation methods enable model migration and adaptation across different domain data distributions. However, the source and target domains of hyperspectral images (HSIs) have large spectral offsets and spatial distribution differences, making the extraction of high-quality domain-invariant features between different domains is essential for classification. To achieve a more consistent feature representation for each category between the source and target domains, we propose a category semantic guided unsupervised domain adaptation network (CSGNet) for HSIs classification. CSGNet is designed to learn cross-domain invariant representation from category semantic information. First, to embed category semantic prior knowledge during feature learning, we extracted textual semantic features from the textual descriptions for each category and projected visual features into the semantic space via visual-linguistic alignment. A category representation memory pool is then introduced to store the visual-linguistic representations of different categories. Second, we propose a bi-classifier adversarial learning method designed to generate inconsistent category predictions in the unlabeled target domain, thereby enhancing the classifier’s discriminative capability regarding those hard-to-transfer features. Finally, to utilize the domain-invariant features stored in the category memory pool, a category attention module is proposed to guide the model’s adaptation to the data from different domains, mitigating the impact of the differences in the domain data distributions. Extensive experimental results validated on three cross-domain datasets demonstrate that the proposed method outperforms other state-of-the-art methods. The source code is available at http://github.com/cuibinge/CSGNet.

Fine-grained classification of coastal wetlands from remote sensing images is a challenging task due to the spectral overlap between different wetland vegetation types, making them difficult to distinguish. Traditional methods for remote sensing interpretation often rely on manual classification or shallow machine learning approaches, which fail to effectively capture complex spatial relationships and contextual information. To integrate ecological and biological prior knowledge and enhance the generalization performance of the model, this paper proposes a remote sensing image classification method for coastal wetlands based on Category Representation Learning, called CRLNet. The core idea is to learn category-invariant representations of land cover types in coastal wetlands using geoscience knowledge graphs and deep neural networks. First, deep feature maps and classification probability maps generated by a semantic segmentation network are used to initialize the representations of each category; then, the Spatial Topological Relationship Encoder (STRE) and Category Attribute Knowledge Encoder (CAKE) are proposed, employing a two-stream architecture to refine the representations of each category; finally, each pixel is assigned to the category with the highest similarity based on the aforementioned deep feature maps and category-invariant representations. By combining graph convolution and self-attention mechanisms, CRLNet effectively integrates ecological and biological prior knowledge into category representation learning, thereby reducing the likelihood of conflicts between classification results and geoscience prior knowledge. Experimental results demonstrate that CRLNet outperforms state-of-the-art methods on the Huanghe River and Yancheng coastal wetland datasets. Notably, CRLNet is a lightweight framework with only 1/55 of the parameter count of CGGLNet, making it computationally efficient while maintaining high classification accuracy. The codes will be available from the website: https://github.com/cuibinge/CRLNet.

Wetland is one of the three ecosystems in the world, and collaborative monitoring using hyperspectral images (HSIs) and light detection and ranging (LiDAR) has been important for wetland ecological protection. However, because of the domain shift of different images, cross-scene wetland classification of HSIs and LiDAR is a practical challenge, necessitating the development of models trained solely on the source domain (SD) and directly transferred to the target domain (TD) without retraining. To address this issue, an instance-wise domain generalization network (IDGnet) is proposed for HSI and LiDAR cross-scene wetland classification. An instance-wise random domain expansion module (IWR-DEM) is developed to simulate the domain shift, establishing the extended domain (ED). Specifically, the original HSI and LiDAR data are separated as semantic and background information in the frequency domain, a random background shift is applied to the HSI, and a semantic random shift is deployed to LiDAR. The HSI and LiDAR fusion features are extracted from the SD and ED by a weight-shared network. Multiple condition constraints are proposed for domain and class alignment, learning the domain-invariant and class-specific information and improving model generalization. Experiments conducted on two wetland datasets demonstrate the superiority of the proposed IDGnet for cross-scene wetland classification with HSI and LiDAR data. The codes will be available from the website:https://github.com/bigshot-g/IEEE_TGRS_IDGnet.

Accurate information on ground features is essential for the ecological protection and efficient management of this wetland. The resolution of satellite remote sensing images often falls short of the requirements for fine-scale classification. The Yellow River Estuary wetland is primarily vegetated, characterized by similar spectral curves. The use of images with limited spectral information additionally restricts the precision of classification. In response to the above issues, this study proposes a multi-branch fusion Transformer (MB Transformer) method based on high spatial resolution unmanned aerial vehicle (UAV) hyperspectral data. By constructing vegetation index features, dimensionality-reduced spectral features, and original spectral features, a multi-branch fusion Transformer model is used to classify in the Yellow River Estuary wetland at a fine scale. The results indicate that this method has obvious advantages in fine-scale classification of wetland in the Yellow River Estuary.

The joint classification of hyperspectral image (HSI) and light detection and ranging (LiDAR) data has attracted considerable attention in the field of remote sensing. Integrating the advantages of the two data sources can provide precise data support and analytical decision-making for remote-sensing applications. However, due to the inherent differences in properties and semantic information from heterogeneous data, most existing deep-learning methods suboptimally extract the characteristic features of both data sources while utilizing their interactive information. In this letter, we propose a dual-branch feature fusion network-based cross-modal enhanced CNN and Transformer (DF2NCECT) to make full use of the respective features and interactive information of multisource data. DF2NCECT consists of two main stages. One is the basic feature extraction stage, which builds a hybrid convolution module based on 3DCNN and inception structure to fully extract the joint features of HSI from multiple spatial perspectives. The other is the deep feature fusion stage, where the CNN and Transformer are designed in parallel to fully explore and fuse deep features between HSI and LiDAR. More importantly, to achieve efficacious interactive information between HSI and LiDAR, a cross-modal enhanced CNN and Transformer module (CECT) is designed to deeply enhance the fused interactive features from global/local perspectives. Experiments show that the proposed method is superior and outperforms the comparison methods by an average of 3.06% in OA on Houston2013 and 1.79% on Summer, respectively.

Coral reef is a typical marine ecosystem and has significant implications for protecting marine biodiversity, and maintaining marine ecological balance. Accurate geomorphic information is the base of coral reef conservation, which usually is extracted by high-resolution remote sensing. Recent classification methods for coral reef geomorphology always focus on the extraction of deep spectral and texture features, ignoring the inherent geospatial information of geomorphology and losing the shallow-layer information, which leads to low classification accuracy. This paper proposes a deep learning classification method for coral reef geomorphology, named as GCU-Net which integrates the convolutional attention mechanism and the geo-spatial cognition. Experiments were carried out in North Reef and Zhaoshu Island geomorphology with the Gaofen-2 (GF-2) satellite image. The results demonstrate that the GCU-Net’s accurate classification with an overall accuracy (OA) of 90.46% and 88.92%, respectively, which better extracts the useful information in the shallow-layer, and effectively reduces the omission and misclassification of geomorphic types due to the different spatial positions. Our method exhibits excellent classification performance with an OA improvement of over 7% compared to the comparison method. Therefore, the method proposed in this paper is more effective in obtaining accurate information on coral reef geomorphology and can provide technical support for carrying out large-scale fine monitoring of coral reefs.

With the development of earth observation technology, hyperspectral image (HSI) and light detection and ranging (LiDAR) data collaborative monitoring has shown great potential in the ecological protection and restoration of coastal wetlands. However, due to the different working principle adopted by the HSI sensor and LiDAR sensor, the data obtained by them has different distribution characteristics. The distribution difference limits the fusion of HSI and LiDAR data, bringing a great challenge for coastal wetland classification. To tackle this problem, a multi-source feature embedding and interaction fusion network is proposed for coastal wetland classification, named MsFE-IFN. First, the HSI and LiDAR data are embedded in the same feature space, where the feature distribution of multi-source remote sensing are aligned to alleviate data distribution differences. Second, the aligned HSI and LiDAR features interact information in channels and pixels, which is able to establish the relationship of spectral, elevation and geospatial. Third, the HSI and LiDAR feature are sent into the feature fusion network, in which the low-frequency residual is retained to enrich intra-class features. Finally, the fused feature is applied for final class prediction. Experiments conducted on three coastal wetland HSI-LiDAR datasets created by ourselves demonstrate the superiority of the proposed MsFE-IFN for coastal wetland classification. The codes will be available from the website:https://github.com/bigshot-g/IEEE_TGRS_MsFE-IFN.

Convolutional neural network (CNN)-based hyperspectral image (HSI) classification models often exhibit high volume and complexity. This not only poses challenges in deploying them on mobile and embedded devices due to storage and power constraints but also introduces a dilemma between the growing demand for labeled samples and the high cost associated with manual labeling. To address these challenges, we propose an ultra-lightweight hybrid CNN based on redundancy removal (ULite-R2HCN), specifically designed for HSI classification in scenarios with limited samples. To reduce computational costs and enhance feature extraction effectiveness, we focus on optimizing the widely used depthwise convolution (DW-Conv) and pointwise convolution (PW-Conv) in the lightweight HSI classification model. For DW-Conv, we design a spatial convolution with redundancy removal (R2Spatial-Conv). This involves the design of multi-scale 3D convolution kernels with specific structures instead of 2D convolution kernels, aiming to reduce redundant convolution kernels and extract multi-scale spatial features. Simultaneously, for PW-Conv, we design a spectral convolution with redundancy removal (R2Spectral-Conv). This utilizes a “copy-splicing-grouping” structure to extract spectral features within arbitrary range intervals, effectively reducing redundant spectral extractions and capturing long-range spectral relationships. Numerous experiments have shown that the proposed ULite-R2HCN achieves higher classification accuracy with an ultra-light volume for a few training samples. In addition, sufficient ablation experiments also verified the advanced performance of the designed R2Spatial-Conv and R2Spectral-Conv.

Few-shot class-incremental learning has recently received significant research focus in remote sensing scene classification (FSCIL-RSSC). The success of FSCIL-RSSC relies on the robustness of the feature backbone and classifiers. Existing works focus on improving classifier adaptation, but little attention is paid to the importance of backbone robustness on the recognition ability of new class samples’ embeddings. Due to the large distribution shift between old and new classes, FSCIL-RSSC using high-layer (single-scale) features may not adapt flawlessly to new categories. To solve the issue, we put forward a gradient guided multiscale feature collaboration network (G-MFCN) for FSCIL-RSSC. Specifically, we introduce a parallel hierarchy strategy to simultaneously capture the multifeature discriminative information of the same sample. Then, a gradient guide block is designed to automatically pick out the optimal values of different convolution blocks for multifeature fusion. Finally, the classical feature pyramid network is introduced for multiscale fusion to obtain more obvious discriminative features of RSSC. More importantly, our proposed G-MFCN is a simple and adaptable module, which can combine any existing FSCIL frameworks to further improve the optimized classifiers’ effectiveness for the FSCIL-RSSC scenario. Extensive experiments on four benchmarks demonstrate that the proposed G-MFCN achieves significant improvements in comparison to existing FSCIL-RSSC methods.

Class-incremental (CI) learning has recently received extensive research interest in remote sensing scene classification (CI-RSSC). The existing CI-RSSC methods’ superior performance seriously relies on old (base classes) and new classes (incremental classes) sampled independently from an identical distribution (dataset). In real-world RSSC scenarios, there exist significant distribution shifts between old and new classes, leading to the existing CI-RSSC methods being unable to adjust flawlessly to these new classes. In this article, we propose a novel cross-domain (CD) CI-RSSC framework to solve the above-mentioned problems, termed CDCI-RSSC. Specifically, a modular sharing-based dynamic extension module is first designed, which only updates specialized modules to extract new class feature embeddings for reducing memory footprint. Then, an effective dynamic alignment guided domain adaptive module (DAM) is further proposed to calculate the dynamic weights of each sample in various fields, which can minimize distribution shifts between source and target domains. Finally, a foreground enhancement module (FEM) is introduced to alleviate the issue of complex background interference in RSSC by increasing the weight of critical regions. Compared with the existing CI-RSSC and CD-RSSC, our proposed CDCI-RSSC framework surmounts the challenge of handling the distribution shifts between source (base session) and target domains (incremental session) while alleviating the limitations of continuous learning of new classes. Extensive experiments on three CDCI scenarios show that the CDCI-RSSC model achieves significant performance improvements in comparison to existing CI-RSSC and CD-RSSC methods.

Recent studies have demonstrated the potential of hybrid convolutional models that combine 3D and 2D convolutional neural networks (CNNs) for hyperspectral image (HSI) classification. However, these models do not fully utilize the benefits of hybrid convolution due to inefficient connections between the two types of CNNs. Moreover, most CNNs, including hybrid models, require a significant number of parameters and computational resources for accurate classification, which increases the need for labeled samples and computational cost. Although the common lightweight strategies like depthwise separable convolution (DSC) can reduce parameters and computation compared to normal convolution (NC), they often compromise accuracy. To address these challenges, we propose a lightweight hybrid convolutional neural network (Lite-HCNet) for HSI classification with minimal model parameters and computational effort. Firstly, we design a novel channel attention module (NCAM) and combine it with a convolutional kernel decomposition (CKD) strategy to propose a lightweight and efficient DSC (LE-DSC) deployed in Lite-HCNet. The LE-DSC not only reduces the DSC volume further but also enhances its performance. Secondly, a lightweight and efficient hybrid convolutional layer (LE-HCL) is designed in Lite-HCNet to explore the efficient connection structure between 3D CNNs and 2D CNNs. Experiments show that the Lite-HCNet reduces the required computational cost and practical deployment difficulty while offering advanced performance with a small number of training samples. Furthermore, abundant ablation experiments confirm the superior performance of the designed LE-DSC.

The fine mapping of coastal wetlands is a major challenge due to the spectral aliasing of vegetation. In this letter, we selected Zhuhai-1 hyperspectral images (HSIs) for coastal wetland mapping and proposed a tiny-scene embedding network (TSE-Net) based on scene representation and attention mechanism. In TSE-Net, the tiny-scene representation associated with each hyperspectral pixel was extracted and used to enhance the spectral discrimination of ground objects. DenseNet was chosen as the backbone network, and the attention mechanism was introduced into the dense blocks to extract remarkable features. Experiments on the Yellow River estuary coastal wetland showed that the results of TSE-Net had a significant improvement in accuracy compared to other models, especially for the coastal wetland vegetation with confusing spectra, such as Spartina alterniflora, Suaeda salsa, Phragmites australis, and Tamarix.

Hyperspectral unmixing (HU) is widely used to process mixed pixels as an essential technology. Among them, the nonnegative matrix factorization (NMF)-based approach is one typical of the blind unmixing techniques, which can achieve endmembers and abundances simultaneously. Considering the physical meaning of the extracted endmembers, the archetypal analysis (AA) method constructs a new matrix decomposition structure with stronger interpretability than NMF. However, AA ignores the significant sparse property of abundance in unmixing. Therefore, we propose the L1sparsity-constrained AA algorithm for HU. To solve the new optimization problem, we explore a new optimization method for optimizing abundance. The alternating direction method of multipliers (ADMM) is used to increase the strong convexity and convergence of the problem. Then the fast gradient method (FGM) instead of traditional gradient descent is used to speed up algorithm convergence. The experimental results in both the synthesized and real datasets show that the proposed method outperforms several sparse NMF-based and AA-based methods.

Microwave remote sensing plays an important role in monitoring the sea surface oil spill, but it has some shortcomings in monitoring oil film thickness. Many scholars have done a lot of work for this. At present, optics, laser and ultraviolet have made some breakthroughs in the study of oil film thickness; In particular, Hyperspectral shows great advantages in oil film thickness (Lu et al. 2011,2013; Brekke and Solberg 2005; Ren et al., 2019).

The phenomenon of spectral aliasing exists for coastal wetland object types, which leads to class mixing. This letter proposes a multiobject convolutional neural network (CNN) decision fusion classification method for hyperspectral images of coastal wetlands. This method adopts decision fusion based on fuzzy membership rules applied to single-object CNN classification to obtain higher classification accuracy. Experimental results demonstrate the effectiveness of the proposed method for the six object types, including water, tidal flat, reed, and other vegetation types. The overall accuracy of the decision fusion classification method based on fuzzy membership is 82.11%, which is 3.33% and 6.24% higher than those of single-object feature band CNN and support vector machine methods. The classification method based on multiobject CNN decision fusion inherits the characteristics of single-object feature bands of the CNN, making it a practical approach to image classification under the challenging conditions in which class mixing occurs.

In this paper, a novel SuperPixel-based Extended Random Walker (SPERW) classification method for hyperspectral images is proposed that consists of three main steps. First, a multiscale segmentation algorithm is adopted to generate many superpixels, each of which represents a homogeneous region of adaptive shape and size. Then, a new weighted graph is constructed based on the superpixels in which the nodes correspond to the superpixels and the edges correspond to the links connecting two adjacent superpixels. Each edge has a weight that defines the similarity between the two superpixels. Second, a widely used pixelwise classifier, i.e., the support vector machine, is adopted to obtain classification probability maps for a hyperspectral image, which are then used to approximate the prior probabilities of the superpixels. Finally, the obtained prior probability maps of the superpixels are optimized by using the Extended Random Walker (ERW) algorithm, which encodes the spatial information both among and within the superpixels of the hyperspectral image in a weighted graph. Compared with the spectrum of a single pixel, the spectrum of a superpixel is more stable and less affected by noise; therefore, superpixels are more appropriate for adoption as the basic elements in the hyperspectral image classification. Because the spectral correlation between pixels within the same superpixel and the spatial correlation among adjacent superpixels are both well considered in the ERW-based global optimization framework, the proposed method shows high classification accuracy on four widely used real hyperspectral data sets even when the number of training samples is relatively small.

Coral resources in global are facing a huge threat under climate change and increasing human activities. In this paper, at first, combined with Support Vector Machine (SVM), WorldView-3 satellite data (October 10, 2014) with high spatial resolution (1.2 m) and eight spectral bands are used to research classification of Zhao Shu island coral reef substrate. Coral reef, coral bleaching, coral sand and seawater are identified. Then training and checking samples are built artificially and label types from in situ data and finished classification result accuracy test. The results show that the total classification accuracy and Kappa coefficient are 95.28% and 0.90 respectively. These results demonstrate that high resolution images can provide more detailed information and are suitable to monitor the health of coral reefs. High resolution remote sensing classification results of coral reef can provide the important information for environment protection.

In this paper, the data source are GF-1 WFV image and Radarsat-2 SAR image covering the Yellow River Estuary wetland eastern area. The paper first uses Gram-Schmidt algorithm for fusing GF-1 image and different polarimetric mode SAR images, and then uses the method of SVM for supervised classification. Finally, the accuracy of the classification results and the capacity of information extraction are compared. The experiment results show:(1) the classification accuracy of fusing the VV polarimetric mode of SAR image and GF-1 image is better than other fusion image, reaching 83.78%, closing to the classification accuracy of GF-1 image. The classification accuracy of tidal flat reed in VV polarimetric fusion image is better than that of GF-1.(2) Tidal flat, river and aquaculture pond have the highest classification accuracy in all the fusion images.