Information Of Hyperspectral Image Research Articles

A graph-guided transformer based on dual-stream perception for hyperspectral image classification

ABSTRACT The excellent capabilities of Transformers and Graph Neural Networks (GNNs) in modelling long-range dependencies and handling irregular data have led to their widespread application in hyperspectral image (HSI) classification tasks. However, the Graph Transformer combining both advantages is rarely used in this field and has some limitations. Current Graph Transformers consider interactions between all nodes within the graph, adding complexity and introducing unnecessary information from noisy nodes. Moreover, the rich spectral information in HSIs is often ignored, and there is a lack of effective fusion of spatial information. In this paper, we propose a dual-stream graph-guided Transformer for HSI classification. In spatial dimension, superpixels are utilized to guide spatial graph generation, capturing global topological dependencies and local details effectively in HSIs. In terms of spectrum, we innovatively construct a spectral graph based on spectral channels and adopt a contribution score-based strategy to adaptively filter out irrelevant edges, achieving sparsity while preserving spectral context relationships. Experimental results demonstrate the significant competitive advantage of our method in HSI classification tasks on three public datasets. The code is available at https://github.com/youngboy03/GTDPNet.

Deep blind super-resolution for hyperspectral images

Current deep learning methods for single hyperspectral image super-resolution are non-blind ones, which adopt the simplistic bicubic degradation model. These models have poor generalization performance and cannot handle unknown degradation. Additionally, blind super-resolution methods for RGB images neglect rich spectral information in hyperspectral images, which causes spectral distortion in results. To address these issues, we consider the degeneration estimation and propose a single hyperspectral image blind super-resolution algorithm. Specifically, we first use a blur kernel estimation network and a deblurring network to obtain the image without blur. We change the receptive field by exchanging spatial channel information when estimating the kernel, thereby obtaining blurry information at different scales. In addition, to reduce the errors introduced by kernel mismatch and the relevant information among bands ignored by the group convolution, we fuse the spatial information of the blurred image to compensate for the errors and then utilize the spatial and spectral information of other bands to guide the feature extraction of the current band. Finally, we integrate the feature information from different branches in a global fusion network to improve spatial and spectral information fidelity further. Extensive experiments on synthetic and real-world hyperspectral datasets unequivocally demonstrate that our approach outperforms other state-of-the-art approaches in terms of both quantitative and qualitative evaluations. Our code is publicly available at https://github.com/YoungP2001/DBSR.

Advanced Global Prototypical Segmentation Framework for Few-Shot Hyperspectral Image Classification.

With the advancement of deep learning, related networks have shown strong performance for Hyperspectral Image (HSI) classification. However, these methods face two main challenges in HSI classification: (1) the inability to capture global information of HSI due to the restriction of patch input and (2) insufficient utilization of information from limited labeled samples. To overcome these challenges, we propose an Advanced Global Prototypical Segmentation (AGPS) framework. Within the AGPS framework, we design a patch-free feature extractor segmentation network (SegNet) based on a fully convolutional network (FCN), which processes the entire HSI to capture global information. To enrich the global information extracted by SegNet, we propose a Fusion of Lateral Connection (FLC) structure that fuses the low-level detailed features of the encoder output with the high-level features of the decoder output. Additionally, we propose an Atrous Spatial Pyramid Pooling-Position Attention (ASPP-PA) module to capture multi-scale spatial positional information. Finally, to explore more valuable information from limited labeled samples, we propose an advanced global prototypical representation learning strategy. Building upon the dual constraints of the global prototypical representation learning strategy, we introduce supervised contrastive learning (CL), which optimizes our network with three different constraints. The experimental results of three public datasets demonstrate that our method outperforms the existing state-of-the-art methods.

LCFE-X: Pluggable Local Channel Feature Extraction model enhancement method for hyperspectral image classification

Spectral information of hyperspectral images offers high-dimensional features for downstream tasks. Local spectral signatures in different wavelength ranges are critical for material identification and pixel classification. In line with conventional vision methods, recent studies have focused on bolstering the spatial modelling capabilities of transformers. However, the focus remains predominantly global when the channel dimension is considered. Current research on enhancing local channel feature extraction is primarily model specific and lacks generalisable methods that can be applied to various models. Therefore, this study introduces a Local Channel Feature Extraction (LCFE) module. It extracts spectral-spatial features and key features within each band group and then integrates the features to obtain local channel features. Subsequently, a pluggable LCFE-X model enhancement method is proposed. This involves performing LCFE at the front end of the model and fusing the extracted features with the original features. Three enhanced hyperspectral image classification models, namely, LCFE-ViT, LCFE-SF, and LCFE-MF, are constructed based on recently proposed models. Extensive experiments demonstrate that the LCFE module effectively extracts local spectral signatures, and the proposed method enhances the performance of the various basic models. The code is available at https://github.com/YiSu1997/LCFE-X.

HyperGCN – a multi-layer multi-exit graph neural network to enhance hyperspectral image classification

ABSTRACT Graph neural networks (GNNs) have recently garnered significant attention due to their exceptional performance across various applications, including hyperspectral (HS) image classification. However, most existing GNN-based models for HS image classification are limited depth models and often suffer from performance degradation as model depth increases. This study introduces HyperGCN, an exclusive GNN-based model designed with multiple graph convolutional layers to exploit the rich spectral information inherent in HS images, thereby enhancing classification performance. To address performance degradation, HyperGCN incorporates techniques resistant to oversmoothing into its architecture. Additionally, multiple-side exit branches are integrated into the intermediate layers of HyperGCN, enabling dynamic management of the complexity of HS images. Less complex HS images are processed by fewer layers, exiting early via attached branches, while more complex images traverse multiple layers until reaching the final output layer. Extensive experiments on four benchmark HS datasets (Indian Pines, Pavia University, Salinas, and Botswana) demonstrate HyperGCN’s superior performance over basic GNN-based models. Notably, HyperGCN outperforms or performs comparably to the CNN-GNN combined model in classifying HS images. Furthermore, the superior performance of multi-exit HyperGCN over its single-exit counterpart emphasizes the effectiveness of incorporating side exit branches in GNN-based HS image classification. Compared to state-of-the-art models, multi-exit HyperGCN demonstrates competitive performance, highlighting its effectiveness in handling complex spectral information in HS images while maintaining an acceptable balance between accuracy and computational efficiency.

Achieving Better Category Separability for Hyperspectral Image Classification: A Spatial-Spectral Approach.

The task of hyperspectral image (HSI) classification has attracted extensive attention. The rich spectral information in HSIs not only provides more detailed information but also brings a lot of redundant information. Redundant information makes spectral curves of different categories have similar trends, which leads to poor category separability. In this article, we achieve better category separability from the perspective of increasing the difference between categories and reducing the variation within category, thus improving the classification accuracy. Specifically, we propose the template spectrum-based processing module from spectral perspective, which can effectively expose the unique characteristics of different categories and reduce the difficulty of model mining key features. Second, we design an adaptive dual attention network from spatial perspective, where the target pixel can adaptively aggregate high-level features by evaluating the confidence of effective information in different receptive fields. Compared with the single adjacency scheme, the adaptive dual attention mechanism makes the ability of target pixel to combine spatial information to reduce variation more stable. Finally, we designed a dispersion loss from the classifier's perspective. By supervising the learnable parameters of the final classification layer, the loss makes the category standard eigenvectors learned by the model more dispersed, which improves the category separability and reduces the rate of misclassification. Experiments on three common datasets show that our proposed method is superior to the comparison method.

Hyperspectral Image Classification Based on Two-Branch Multiscale Spatial Spectral Feature Fusion with Self-Attention Mechanisms

In recent years, the use of deep neural network in effective network feature extraction and the design of efficient and high-precision hyperspectral image classification algorithms has gradually become a research hotspot for scholars. However, due to the difficulty of obtaining hyperspectral images and the high cost of annotation, the training samples are very limited. In order to cope with the small sample problem, researchers often deepen the network model and use the attention mechanism to extract features; however, as the network model continues to deepen, the gradient disappears, the feature extraction ability is insufficient, and the computational cost is high. Therefore, how to make full use of the spectral and spatial information in limited samples has gradually become a difficult problem. In order to cope with such problems, this paper proposes two-branch multiscale spatial–spectral feature aggregation with a self-attention mechanism for a hyperspectral image classification model (FHDANet); the model constructs a dense two-branch pyramid structure, which can achieve the high efficiency extraction of joint spatial–spectral feature information and spectral feature information, reduce feature loss to a large extent, and strengthen the model’s ability to extract contextual information. A channel–space attention module, ECBAM, is proposed, which greatly improves the extraction ability of the model for salient features, and a spatial information extraction module based on the deep feature fusion strategy HLDFF is proposed, which fully strengthens feature reusability and mitigates the feature loss problem brought about by the deepening of the model. Compared with five hyperspectral image classification algorithms, SVM, SSRN, A2S2K-ResNet, HyBridSN, SSDGL, RSSGL and LANet, this method significantly improves the classification performance on four representative datasets. Experiments have demonstrated that FHDANet can better extract and utilise the spatial and spectral information in hyperspectral images with excellent classification performance under small sample conditions.

A joint model based on graph and deep learning for hyperspectral anomaly detection

Through the years, graph theory has gradually been applied in hyperspectral image (HSI) processing. The graph theory method does not need to consider the structural characteristics of the dataset and has achieved satisfactory application results. However, the current graph theory methods applied in HSI processing mainly focus on spectral characteristics. This paper proposes a joint model based on graph and deep learning (JGD) for hyperspectral anomaly detection (AD). The proposed JGD consists of a spectral sub-model using graph and a spatial sub-model using deep learning. In the spectral sub-model, graph Fourier transform (GFT) is used, and the reconstruction errors between the original HSI and the HSI after GFT are mapped to fractional Fourier domain (FrFD) for AD results, which mainly makes use of the global spectral information. In the spatial sub-model, stacked autoencoders (SAEs) and an adaptive algorithm are combined to obtain AD results, which mainly use local spatial information and spectral information. These two sub-models are combined through the differential fusion method to form the JGD, which fully utilizes spectral and spatial information of hyperspectral images (HSIs). Compared with seven other algorithms on four real HSIs, the proposed JGD shows superior AD performance.

Unsupervised hyperspectral images classification using hypergraph convolutional extreme learning machines

AbstractAiming at the problem that traditional methods are difficult to fully utilize the rich spectral information in hyperspectral images (HSI) and fail to capture the complex higher‐order relations in hyperspectral data, which leads to limited classification performance extreme learning machine and fails to further improve the classification accuracy of HSIs, the authors propose the hypergraph convolutional extreme learning machine (HGCELM) method. The method not only inherits all the advantages of extreme learning machine (ELM), but also embeds hypergraph convolution for feature selection, which is capable of handling higher‐order relations. This enables HGCELM to capture more complex relationships between nodes and provide richer representation capabilities. At the same time, the training speed advantage of ELM is retained, thus speeding up the model training process. Experimental results show that the proposed algorithm achieves better accuracy compared to other clustering algorithms.

Joint Classification of Hyperspectral and LiDAR Data Based on Adaptive Gating Mechanism and Learnable Transformer

Utilizing multi-modal data, as opposed to only hyperspectral image (HSI), enhances target identification accuracy in remote sensing. Transformers are applied to multi-modal data classification for their long-range dependency but often overlook intrinsic image structure by directly flattening image blocks into vectors. Moreover, as the encoder deepens, unprofitable information negatively impacts classification performance. Therefore, this paper proposes a learnable transformer with an adaptive gating mechanism (AGMLT). Firstly, a spectral–spatial adaptive gating mechanism (SSAGM) is designed to comprehensively extract the local information from images. It mainly contains point depthwise attention (PDWA) and asymmetric depthwise attention (ADWA). The former is for extracting spectral information of HSI, and the latter is for extracting spatial information of HSI and elevation information of LiDAR-derived rasterized digital surface models (LiDAR-DSM). By omitting linear layers, local continuity is maintained. Then, the layer Scale and learnable transition matrix are introduced to the original transformer encoder and self-attention to form the learnable transformer (L-Former). It improves data dynamics and prevents performance degradation as the encoder deepens. Subsequently, learnable cross-attention (LC-Attention) with the learnable transfer matrix is designed to augment the fusion of multi-modal data by enriching feature information. Finally, poly loss, known for its adaptability with multi-modal data, is employed in training the model. Experiments in the paper are conducted on four famous multi-modal datasets: Trento (TR), MUUFL (MU), Augsburg (AU), and Houston2013 (HU). The results show that AGMLT achieves optimal performance over some existing models.

Enhanced deep unrolling networks for snapshot compressive hyperspectral imaging

Snapshot compressive hyperspectral imaging necessitates the reconstruction of a complete hyperspectral image from its compressive snapshot measurement, presenting a challenging inverse problem. This paper proposes an enhanced deep unrolling neural network, called EDUNet, to tackle this problem. The EDUNet is constructed via the deep unrolling of a proximal gradient descent algorithm and introduces two innovative modules for gradient-driven update and proximal mapping reflectivity. The gradient-driven update module leverages a memory-assistant descent approach inspired by momentum-based acceleration techniques, for enhancing the unrolled reconstruction process and improving convergence. The proximal mapping is modeled by a sub-network with a cross-stage spectral self-attention, which effectively exploits the inherent self-similarities present in hyperspectral images along the spectral axis. It also enhances feature flow throughout the network, contributing to reconstruction performance gain. Furthermore, we introduce a spectral geometry consistency loss, encouraging EDUNet to prioritize the geometric layouts of spectral curves, leading to a more precise capture of spectral information in hyperspectral images. Experiments are conducted using three benchmark datasets including KAIST, ICVL, and Harvard, along with some real data, comprising a total of 73 samples. The experimental results demonstrate that EDUNet outperforms 15 competing models across four metrics including PSNR, SSIM, SAM, and ERGAS.

Applying dimensionality reduction methods to extract physiological and diagnostic features for clinical Hyperspectral Images

The most valuable information of Hyperspectral Image (HSI) should be processed properly. Using dimensionality reduction techniques in two distinct approaches, we created a structure for HSI to collect physiological and diagnostic information. The tissue Oxygen Saturation Level (StO2) was extracted using the HSI approach as a physiological characteristic for stress detection. Our research findings suggest that this unique characteristic may not be affected by humidity or temperature in the environment. Comparing the standard StO2 reference and pressure concentrations, the social stress assessments showed a substantial variance and considerable practical differentiation. The proposed system has already been evaluated on tumor images from rats with head and neck cancers using a spectrum from 450 to 900 nm wavelength. The Fourier transformation was developed to improve precision, and normalize the brightness and mean spectrum components. The analysis of results showed that in a difficult situation where awareness could be inexpensive due to feature possibilities for rapid classification tasks and significant in measuring the structure of HSI analysis for cancer detection throughout the surgical resection of wildlife. Our proposed model improves performance measures such as reliability at 89.62% and accuracy at 95.26% when compared with existing systems.

S[formula omitted]WaveNet: A novel spectral–spatial wave network for hyperspectral image classification

Deep learning has made significant progress in hyperspectral image (HSI) classification, and its powerful ability to automatically learn abstract features is well recognized. Recently, the simple architecture of multi-layer perceptron (MLP) has been extensively employed to extract long-range dependencies of HSI and achieved impressive results. However, existing MLP-based models exhibit insufficient representation of spectral–spatial information in HSI and generally aggregate features with fixed weights, which limits their ability to capture semantic differences. To tackle these challenges, this paper proposes a novel spectral–spatial wave network (S2WaveNet) for HSI classification tasks to enhance the representation capability of spectral–spatial features in ground objects. Specifically, the spectral–spatial wave mixer (S2WaveMixer) block is designed as a key component to represent each HSI input as a wave function with amplitude and phase parts. Thus, it enables a deeper dynamic perception and facilitates the extraction of spectral–spatial feature variations of ground objects. The amplitude represents the original features and the phase term is a complex value changing based on the semantic contents of the input images. Furthermore, the inception unit is introduced into the S2WaveMixer block to consider spectral–spatial information at multiple granularity levels. Experiments conducted on five public datasets demonstrate the superiority of S2WaveNet in classification performance and generalization compared to competitors.

ATSFCNN: a novel attention-based triple-stream fused CNN model for hyperspectral image classification

Abstract Recently, the convolutional neural network (CNN) has gained increasing importance in hyperspectral image (HSI) classification thanks to its superior performance. However, most of the previous research has mainly focused on 2D-CNN, and the limited applications of 3D-CNN have been attributed to its complexity, despite its potential to enhance information extraction between adjacent channels of the image. Moreover, 1D-CNN is typically restricted to the field of signal processing as it ignores the spatial information of HSIs. In this paper, we propose a novel CNN model named attention-based triple-stream fused CNN (ATSFCNN) that fuses the features of 1D-CNN, 2D-CNN, and 3D-CNN to consider all the relevant information of the hyperspectral dataset. Our contributions are twofold: First, we propose a strategy to extract and homogenize features from 1D, 2D, and 3D CNN. Secondly, we propose a way to efficiently fuse these features. This attention-based methodology adeptly integrates features from the triple streams, thereby transcending the former limitations of singular stream utilization. Consequently, it becomes capable of attaining elevated outcomes in the context of hyperspectral classification, marked by increased levels of both accuracy and stability. We compared the results of ATSFCNN with those of other deep learning models, including 1D-CNN, 2D-CNN, 2D-CNN+PCA, 3D-CNN, and 3D-CNN+PCA, and demonstrated its superior performance and robustness. Quantitative assessments, predicated on the metrics of overall accuracy (OA), average accuracy (AA), and kappa coefficient (κ) emphatically corroborate the preeminence of ATSFCNN. Notably, spanning the three remote sensing datasets, ATSFCNN consistently achieves peak levels of OA, quantified at 98.38%, 97.09%, and 96.93% respectively. This prowess is further accentuated by concomitant AA scores of 98.47%, 95.80%, and 95.80%, as well as kappa coefficient values amounting to 97.41%, 96.14%, and 95.21%.

Multigraph approximate-representation learning for hyperspectral band selection

ABSTRACT Unsupervised hyperspectral image (HSI) band selection methods have been attracting ever-increasing attention. However, the local structural features captured by most of the existing methods suffer from a certain degree of risk of feature ambiguity. Moreover, these methods typically have difficulty preserving the latent manifold structure of the HSI in selected subsets of the bands. To address these problems, firstly, the graph approximate representation learning (GARL) model is proposed in this article by incorporating graph-regularized sparse coding into self-representation learning to preserve the local manifold structure information of HSI in the selected subset of bands. Meanwhile, to alleviate the ambiguity of the local graph structures, we design a local-global feature relationship expression model to construct the latent long-distance contextual connectivity (LDCC) graphs between the obtained local graph structures. Then, to preserve HSI’s local and non-local manifold structure information as much as possible in the selected subset of bands, a multigraph approximate representation learning (MGARL) model is proposed in this article by incorporating the obtained LDCC graph into the proposed GARL model. Next, we design a solution method to solve the proposed formulation. Finally, extensive experimental results on four datasets reveal the promising performance of MGARL over some state-of-the-art methods.

CrossU-Net: Dual-modality cross-attention U-Net for segmentation of precancerous lesions in gastric cancer

Gastric precancerous lesions (GPL) significantly elevate the risk of gastric cancer, and precise diagnosis and timely intervention are critical for patient survival. Due to the elusive pathological features of precancerous lesions, the early detection rate is less than 10%, which hinders lesion localization and diagnosis. In this paper, we provide a GPL pathological dataset and propose a novel method for improving the segmentation accuracy on a limited-scale dataset, namely RGB and Hyperspectral dual-modal pathological image Cross-attention U-Net (CrossU-Net). Specifically, we present a self-supervised pre-training model for hyperspectral images to serve downstream segmentation tasks. Secondly, we design a dual-stream U-Net-based network to extract features from different modal images. To promote information exchange between spatial information in RGB images and spectral information in hyperspectral images, we customize the cross-attention mechanism between the two networks. Furthermore, we use an intermediate agent in this mechanism to improve computational efficiency. Finally, we add a distillation loss to align predicted results for both branches, improving network generalization. Experimental results show that our CrossU-Net achieves accuracy and Dice of 96.53% and 91.62%, respectively, for GPL lesion segmentation, providing a promising spectral research approach for the localization and subsequent quantitative analysis of pathological features in early diagnosis.

A hyperspectral band selection algorithm for identifying high oleic acid peanuts

High oleic acid peanuts have higher oleic acid content and stronger oxidation stability than common peanuts, but their appearances are similar, which imposes difficulties for classifying. Based on this, the study aims to classify high oleic acid peanut to ensure its purity by using hyperspectral imaging technology. However, classification accuracy and efficiency are limited given the large amount of redundant information of hyperspectral images. The band iteration algorithm (BIA) is proposed to select characteristic bands by reducing the redundant information between spectral bands for the peanut classification. Hyperspectral images with 616 bands (from 400 nm to 1100 nm) of 126 high oleic acid peanuts and 126 common peanuts were collected. Then, BIA selected optimal bands as characteristic bands from adjacent bands according to the classification accuracy of each band subsets. Thirdly, three classification models, namely linear discriminant analysis, support vector machine, and partial least squares-discriminant analysis (PLS-DA), were employed to compare the performance of BIA with successive projections algorithm and competitive adaptive reweighted sampling, respectively. The experimental results show that BIA can effectively improve the classification ability of spectral data. The BIA-PLS-DA model had the best classification efficiency, and the accuracy of the test set reached 93.26%. For peanut individuals, only one peanut sample was misclassified with a classification error rate of 1.43%.

SemanticFormer: Hyperspectral image classification via semantic transformer

Hyperspectral image (HSI) classification is an active research problem in computer vision and multimedia field. Contrary to traditional image data, HSIs contain rich spectral, spatial and semantic information. Thus, how to extract the discriminative features for HSIs by integrating spectral, spatial and semantic cues together is the core issue to address HSI classification task. Existing works mainly focus on exploring spectral and spatial information which usually fail to fully explore the rich semantic information in HSIs. To address this issue, in this paper, we first propose a novel semantic Transformer scheme, named SemanticFormer, which aims to learn discriminative visual representations for semantics by exploiting the interaction among different semantic tokens. Using the proposed SemanticFormer, we then propose a novel heterogeneous network that contains both spectral–spatial convolution network branch and SemanticFormer branch to extract spectral–spatial and semantic features simultaneously for HSIs. Experiments on two widely used datasets demonstrate the effectiveness of our SemanticFormer and HSI classification network method. Our codes will be available in https://github.com/SissiW/SemanticFormer.

Removal of Mixed Noise in Hyperspectral Images Based on Subspace Representation and Nonlocal Low-Rank Tensor Decomposition.

Hyperspectral images (HSIs) contain abundant spectral and spatial structural information, but they are inevitably contaminated by a variety of noises during data reception and transmission, leading to image quality degradation and subsequent application hindrance. Hence, removing mixed noise from hyperspectral images is an important step in improving the performance of subsequent image processing. It is a well-established fact that the data information of hyperspectral images can be effectively represented by a global spectral low-rank subspace due to the high redundancy and correlation (RAC) in the spatial and spectral domains. Taking advantage of this property, a new algorithm based on subspace representation and nonlocal low-rank tensor decomposition is proposed to filter the mixed noise of hyperspectral images. The algorithm first obtains the subspace representation of the hyperspectral image by utilizing the spectral low-rank property and obtains the orthogonal basis and representation coefficient image (RCI). Then, the representation coefficient image is grouped and denoised using tensor decomposition and wavelet decomposition, respectively, according to the spatial nonlocal self-similarity. Afterward, the orthogonal basis and denoised representation coefficient image are optimized using the alternating direction method of multipliers (ADMM). Finally, iterative regularization is used to update the image to obtain the final denoised hyperspectral image. Experiments on both simulated and real datasets demonstrate that the algorithm proposed in this paper is superior to related mainstream methods in both quantitative metrics and intuitive vision. Because it is denoising for image subspace, the time complexity is greatly reduced and is lower than related denoising algorithms in terms of computational cost.

Exploring the Spectral Prior for Hyperspectral Image Super-Resolution.

In recent years, many single hyperspectral image super-resolution methods have emerged to enhance the spatial resolution of hyperspectral images without hardware modification. However, existing methods typically face two significant challenges. First, they struggle to handle the high-dimensional nature of hyperspectral data, which often results in high computational complexity and inefficient information utilization. Second, they have not fully leveraged the abundant spectral information in hyperspectral images. To address these challenges, we propose a novel hyperspectral super-resolution network named SNLSR, which transfers the super-resolution problem into the abundance domain. Our SNLSR leverages a spatial preserve decomposition network to estimate the abundance representations of the input hyperspectral image. Notably, the network acknowledges and utilizes the commonly overlooked spatial correlations of hyperspectral images, leading to better reconstruction performance. Then, the estimated low-resolution abundance is super-resolved through a spatial spectral attention network, where the informative features from both spatial and spectral domains are fully exploited. Considering that the hyperspectral image is highly spectrally correlated, we customize a spectral-wise non-local attention module to mine similar pixels along spectral dimension for high-frequency detail recovery. Extensive experiments demonstrate the superiority of our method over other state-of-the-art methods both visually and metrically. Our code is publicly available at https://github.com/HuQ1an/SNLSR.