Hyperspectral image denoising based on multi-resolution dense memory network

被引:2
|
作者
Li, Kengpeng [1 ]
Qi, Jinli [1 ]
Sun, Lei [1 ]
机构
[1] Sun Yat Sen Univ, Sch Syst Sci & Engn, Guangzhou, Guangdong, Peoples R China
关键词
Hyperspectral image (HSI) denoising; Multi-resolution; Power law memory; Densely connected convolutional networks; RESTORATION;
D O I
10.1007/s11042-023-14778-3
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Hyperspectral images (HSIs) denoising is an important pre-processing step since noise will seriously degrade the HSIs quality. In this paper, we propose a multi-resolution dense memory denoising network. Specifically, in order to fully use the spatial-spectral correlation of hyperspectral image (HSI), target noisy band and multiple adjacent bands of HSI are extracted as our network input. After feature extraction by convolution kernels, the feature map is divided into resolution of different sizes through average pooling operation. For each resolution, according to the power law distribution of memory system, we design a dense connection structure in which we use weighted sum of the output feature maps for all previous layers. Then, the deconvolution is used for up sampling. Finally, dilated convolution and skip connection are utilized to obtain the denoised HSI. Our method outperforms state-of-the-art denoising algorithms in both quantitative metrics (e.g. PSNR and SSIM) and visual effects on simulated datasets. The real datasets (Indian Pines) experiment demonstrates that the classification accuracy of the denoised HSI is improved by 20.26% compared with the noisy HSI.
引用
收藏
页码:29733 / 29752
页数:20
相关论文
共 50 条
  • [41] Robust Content Based Image Retrieval Based On Multi-Resolution Wavelet Features and Edge Histogram
    Patvardhan, C.
    Verma, A. K.
    Lakshmi, C. Vasantha
    2013 IEEE SECOND INTERNATIONAL CONFERENCE ON IMAGE INFORMATION PROCESSING (ICIIP), 2013, : 447 - 452
  • [42] Nonlocal Spatial-Spectral Neural Network for Hyperspectral Image Denoising
    Fu, Guanyiman
    Xiong, Fengchao
    Lu, Jianfeng
    Zhou, Jun
    Qian, Yuntao
    IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2022, 60
  • [43] Underwater Image Enhancement Based on Luminance Reconstruction by Multi-Resolution Fusion of RGB Channels
    Wang, Yi
    Chen, Zhihua
    Yan, Guoxu
    Zhang, Jiarui
    Hu, Bo
    SENSORS, 2024, 24 (17)
  • [44] Image Registration Method Based on Multi-resolution for Dual-Energy Subtraction Radiography
    Kawamura, Takahiro
    Omae, Norihiro
    Yamada, Masahiko
    Ito, Wataru
    Kawamoto, Kiyosumi
    Doi, Tsukasa
    MEDICAL IMAGING 2012: IMAGE PROCESSING, 2012, 8314
  • [45] Multi-resolution texture synthesis from turntable image sequences
    Wang, Xuedong
    Wu, Xiaojun
    Zhang, Xiaorong
    FRONTIERS OF MECHANICAL ENGINEERING, 2012, 7 (01) : 72 - 80
  • [46] Anatomy-guided multi-resolution image reconstruction in PET
    Lesonen, P.
    Wettenhovi, V-v
    Kolehmainen, V
    Pulkkinen, A.
    Vauhkonen, M.
    PHYSICS IN MEDICINE AND BIOLOGY, 2024, 69 (10):
  • [47] SALIENT REGION DETECTION BASED ON MULTI-RESOLUTION
    Guo, Ying-Chun
    Yue, Xiao-Min
    Yan, Gang
    PROCEEDINGS OF 2013 INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND CYBERNETICS (ICMLC), VOLS 1-4, 2013, : 968 - 972
  • [48] Multi-resolution texture synthesis from turntable image sequences
    Xuedong Wang
    Xiaojun Wu
    Xiaorong Zhang
    Frontiers of Mechanical Engineering, 2012, 7 (1) : 72 - 80
  • [49] MULTI-RESOLUTION SPATIAL UNMIXING FOR MERIS AND LANDSAT IMAGE FUSION
    Amoros-Lopez, J.
    Gomez-Chova, L.
    Guanter, L.
    Alonso, L.
    Moreno, J.
    Camps-Valls, G.
    2010 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, 2010, : 3672 - 3675
  • [50] An efficient multi-resolution GA approach to dental image alignment
    Nassar, Diaa Eldin
    Ogirala, Mythili
    Adjeroh, Donald
    Ammar, Hany
    IMAGE PROCESSING: ALGORITHMS AND SYSTEMS, NEURAL NETWORKS, AND MACHINE LEARNING, 2006, 6064