计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (12): 189-199.doi: 10.11896/jsjkx.250100082

• 计算机图形学&多媒体 • 上一篇    下一篇

面向高光谱图像去噪的超像素级图特征学习方法

吴颖1, 叶海良1, 曹飞龙2   

  1. 1 中国计量大学理学院 杭州 310018
    2 浙江师范大学数学科学学院 浙江 金华 321014
  • 收稿日期:2025-01-13 修回日期:2025-04-23 出版日期:2025-12-15 发布日期:2025-12-09
  • 通讯作者: 曹飞龙(caofeilong88@zjnu.edu.cn)
  • 作者简介:(wuying991105@163.com)
  • 基金资助:
    国家自然科学基金面上项目(62176244,62536006)

Superpixel-level Graph Feature Learning Method for Hyperspectral Image Denoising

WU Ying1, YE Hailiang1, CAO Feilong2   

  1. 1 School of Sciences, China Jiliang University, Hangzhou 310018, China
    2 School of Mathematical Sciences, Zhejiang Normal University, Jinhua, Zhejiang 321014, China
  • Received:2025-01-13 Revised:2025-04-23 Published:2025-12-15 Online:2025-12-09
  • About author:WU Ying,born in 1999,master.Her main research interests include deep learning and hyperspectral image denoising.
    CAO Feilong,born in 1965,Ph.D,professor,Ph.D supervisor.His main research interests include deep learning and image processing.
  • Supported by:
    This work was supported by the National Natural Science Foundation of China(62176244,62536006).

PDF (PC)

9

摘要： 基于传统深度学习的高光谱图像去噪方法通常难以捕捉空间位置的长程相关性和全局不规则局部块的相似性,导致去噪后细节信息丢失和结构完整性不足。为此,提出一种新的面向高光谱图像去噪的超像素级图特征学习方法,旨在利用图神经网络提取空谱特征,捕捉不规则局部块空间位置的长程相关性,以保留纹理细节和结构信息。首先,设计了门控注意力模块来抑制噪声并增强光谱相关性,为后续的超像素分割奠定基础。然后,设计了超像素级图聚合模块,将高光谱图像按空间维分割成多个空间相连的超像素,并使用共享线性层学习超像素中像素的加权值。接着,使用图卷积更新节点信息,从而有效地保持高光谱图像结构的完整性和内部细节的清晰度。最后,利用高光谱图像的低秩性,在训练过程中引入核范数正则化进行约束,提出了低秩-空谱损失,以关注结构信息的保留。实验结果表明,所提方法在性能上优于当前先进方法。

关键词: 高光谱图像去噪, 图神经网络, 深度学习, 空谱特征, 核范数

Abstract: Hyperspectral image denoising methods based on traditional deep learning usually have difficulty capturing the long-range correlation of spatial positions and the similarity of global irregular local blocks,resulting in loss of detailed information and insufficient structural integrity after denoising.To this end,this paper proposes a new superpixel-level graph feature learning method for hyperspectral image denoising,which aims to use graph neural networks to extract spatial-spectral features and capture the long-range correlation of spatial positions of irregular local blocks to preserve texture details and structural information.Firstly,a gated attention module is designed to suppress noise and enhance spectral correlation,laying the foundation for subsequent superpixel segmentation.Then,a superpixel-level graph aggregation module is designed,which effectively maintains the structural integrity and clarity of internal details of the hyperspectral image by segmenting the hyperspectral image into multiple spatially connected superpixels according to the spatial dimensions and using a shared linear layer to learn the weighted values of pixels in the superpixel.Then,graph convolution is used to update the node information.Finally,nuclear norm regularization is introduced in the training process for constraint by taking advantage of the low rank of hyperspectral images,and a low-rank-spatial spectral denoising loss is proposed to focus on preserving structural information.Experimental results show that the proposed method outperforms the current advanced methods in performance.

Key words: Hyperspectral image denoising, Graph neural networks, Deep learning, Spatial-spectral features, Nuclear norm

中图分类号: 

  • TP751
[1]SUN S,LIU J,LI W.Spatial Invariant Tensor Self-Representation Model for Hyperspectral Anomaly Detection [J].IEEE Transactions on Cybernetics,2024,54(5):3120-3131.
[2]LYU Z Y,ZHANG M,SUN W W,et al.Spatial-Contextual Information Utilization Framework for Land Cover Change Detection with Hyperspectral Remote Sensed Images [J].IEEE Transactions on Geoscience and Remote Sensing,2023,61:4411911.
[3]BEGLIOMINI F N,BARBOSA C C F,MARTINS V S,et al.Machine Learning for Cyanobacteria Mapping on Tropical Urban Reservoirs Using PRISMA Hyperspectral Data [J].ISPRS Journal of Photogrammetry and Remote Sensing,2023,204:378-396.
[4]ZHUANG L,NG M K.FastHyMix:Fast and Parameter-FreeHyperspectral Image Mixed Noise Removal [J].IEEE Transactions on Neural Networks and Learning Systems,2023,34(8):4702-4716.
[5]ZHANG Q,ZHENG Y M,YUAN Q Q,et al.HyperspectralImage Denoising:From Model-Driven,Data-Driven,to Model-Data-Driven [J].IEEE Transactions on Neural Networks and Learning Systems,2024,35(10):13143-13163.
[6]CHEN Y R,ZHANG H,WANG Y N,et al.Flex-DLD:Deep Low-Rank Decomposition Model with Flexible Priors for Hyperspectral Image Denoising and Restoration [J].IEEE Transactions on Image Processing,2024,33:1211-1226.
[7]ZHENG J W,HUANG J J,QIN M J,et al.Hyperspectral Image Denoising Based on Non-Local Similarity and Weighted-Truncated Nuclearnorm [J].Computer Science,2021,48(9):160-167.
[8]HE W,YAO Q M,LI C,et al.Non-Local Meets Global:An Ite-rative Paradigm for Hyperspectral Image Restoration [J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2022,44(4):2089-2107.
[9]ZHANG L H,YE J.Hyperspectral Image Denoising Based onGroup Sparse and Constraint Smooth Rank Approximation [J].Computer Science,2023,50(6):209-215.
[10]SU X,ZHANG Z,YANG F.Fast Hyperspectral Image Denoi-sing and Destriping Method Based on Graph Laplacian Regularization [J].IEEE Transactions on Geoscience Remote Sensing,2023,61:5511214.
[11]SI W N,YE J,JIANG B.Hyperspectral Image Denoising Combining Group Sparse and Representative Coefficient Bidirectional Spatial Spectral Total Variation [J].Computer Science,2024,51(12):199-208.
[12]ZHA Z Y,WEN B H,YUAN X,et al.Nonlocal StructuredSparsity Regularization Modeling for Hyperspectral Image Denoising [J].IEEE Transactions on Geoscience and Remote Sensing,2023,61:5510316.
[13]YE H L,LI H,YANG B,et al.A Novel Rank Approximation Method for Mixture Noise Removal of Hyperspectral Images [J].IEEE Transactions on Geoscience and Remote Sensing,2019,57(7):4457-4469.
[14]ZHANG F,ZHANG K,WAN W B,et al.3D Geometrical Total Variation Regularized Low-Rank Matrix Factorization for Hyperspectral Image Denoising [J].Signal Processing,2023,207:108942.
[15]XIE W Y,LI Y S.Hyperspectral Imagery Denoising by Deep Learning with Trainable Nonlinearity Function [J].IEEE Geoscience and Remote Sensing Letters,2017,14(11):1963-1967.
[16]YUAN Q Q,ZHANG Q,LI J,et al.Hyperspectral Image Denoising Employing a Spatial-Spectral Deep Residual Convolutional Neural Network [J].IEEE Transactions on Geoscience and Remote Sensing,2019,57(2):1205-1218.
[17]DONG W S,WANG H,WU F F,et al.Deep Spatial-SpectralRepresentation Learning for Hyperspectral Image Denoising [J].IEEE Transactions on Computational Imaging,2019,5(4):635-648.
[18]MA H W,LIU G C,YUAN Y.Enhanced Non-Local Cascading Network with Attention Mechanism for Hyperspectral Image Denoising [C]//Proceedings of the IEEE International Confe-rence on Acoustics,Speech and Signal Processing.IEEE,2020:2448-2452.
[19]KAN Z W,LI S H,HOU M Z,et al.Attention-Based Octave Network for Hyperspectral Image Denoising [J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2022,15:1089-1102.
[20]TORUN O,YUKSEL S E,ERDEM E,et al.HyperspectralImage Denoising via Self-Modulating Convolutional Neural Networks [J].Signal Processing,2024,214:109248.
[21]PAN E T,MA Y,MEI X G,et al.Hyperspectral Image Denoi-sing via Spectral Noise Distribution Bootstrap [J].Pattern Re-cognition,2023,142:109699.
[22]VASWANI A,SHAZEER N,PARMAR N,et al.Attention Is All You Need [C]//Advances in Neural Information Processing Systems.MIT,2017:5998-6008.
[23]LI M Y,FU Y,ZHANG Y L.Spatial-Spectral Transformer for Hyperspectral Image Denoising [C]//Proceedings of the AAAI Conference on Artificial Intelligence.AAAI,2023:1368-1376.
[24]LAI Z Q,YAN C G,FU Y.Hybrid Spectral Denoising Transformer with Guided Attention [C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.IEEE,2023:13065-13075.
[25]ZHANG Q,YUAN Q Q,LI J,et al.Hybrid Noise Removal in Hyperspectral Imagery with a Spatial-Spectral Gradient Network [J].IEEE Transactions on Geoscience and Remote Sen-sing,2019,57(10):7317-7329.
[26]WANG M Y,HE W,ZHANG S H Y.A Spatial-Spectral Transformer Network with Total Variation Loss for Hyperspectral Image Denoising [J].IEEE Geoscience and Remote Sensing Letters,2023,20:5503105.
[27]LI M,ZHANG L,CUI L X,et al.BLoG:Bootstrapped Graph Representation Learning with Local and Global Regularization for Recommendation [J].Pattern Recognition,2023,144:109874.
[28]JIANG Q T,YE H L,YANG B,et al.Label-Decoupled Medical Image Segmentation with Spatial-Channel Graph Convolution and Dual Attention Enhancement [J].IEEE Journal of Biome-dical and Health Informatics,2024,28(5):2830-2841.
[29]DONG Y N,LIU Q W,DU B,et al.Weighted Feature Fusion of Convolutional Neural Network and Graph Attention Network for Hyperspectral Image Classification [J].IEEE Transactions on Image Processing,2022,31:1559-1572.
[30]YANG F,CHEN X,ZHANG Z,et al.Denoising and Destriping Hyperspectral Images Using Double Graph Laplacian Regularizers [J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2023,16:10406-10419.
[31]WANG P Q,CHEN P F,YUAN Y,et al.Understanding Convolution for Semantic Segmentation [C]//Proceedings of the IEEE Winter Conference on Applications of Computer Vision.IEEE,2018:1451-1460.
[32]ACHANTA R,SHAJI A,SMITH K,et al.SLIC SuperpixelsCompared to State-of-the-Art Superpixel Methods [J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2012,34(11):2274-2282.
[33]CHEN Y P,DAI X Y,LIU M C,et al.Dynamic Convolution:Attention over Convolution Kernels [C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.IEEE,2020:11030-11039.
[34]WOO S,PARK J,LEE J Y,et al.CBAM:Convolutional Block Attention Module [C]//Proceedings of the European Confe-rence on Computer Vision.Springer,2018:3-19.
[35]VERMA S,SHARMA A,SHESHADRI R,et al.GraphFill:Deep Image Inpainting Using Graphs [C]//Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision.IEEE,2024:4996-5006.
[36]JAMPANI V,SUN D,LIU M Y,et al.Superpixel SamplingNetworks [C]//Proceedings of the European Conference on Computer Vision.Springer,2018:352-368.
[37]GLOROT X,BENGIO Y.Understanding the Difficulty of Trai-ning Deep Feedforward Neural Networks [C]//Proceedings of the International Conference on Artificial Intelligence and Statistics.2010:249-256.
[38]KINGMA D P,BA L J.Adam:A Method for Stochastic Optimization [C]//International Conference on Learning Representations.San Diego,USA:ArXiv,2015.
[39]XIONG F C,ZHOU J,ZHAO Q L,et al.MAC-Net:Model-Aided Nonlocal Neural Network for Hyperspectral Image Denoising [J].IEEE Transactions on Geoscience and Remote Sensing,2022,60:5519414.
[40]ZHU X,MILANFAR P.Automatic Parameter Selection for Denoising Algorithms Using a No-Reference Measure of Image Content [J].IEEE Transactions on Image Processing,2010,19(12):3116-3132.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发