Acta Optica Sinica, Volume. 44, Issue 13, 1310001(2024)

Self-Supervised Enhancement of Low-Light Images Based on Blind Spot Networks

Yong Chen1、*, Jinliang Zhang1, Huanlin Liu2, Kaixin Shao1, Shangming Chen1, Hangying Xiong2, and Yourui Zhang1
Author Affiliations
  • 1Key Laboratory of Industrial Internet of Things and Network Control, Ministry of Education, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
  • 2School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
  • show less
    Figures & Tables(13)
    Overall framework of self-supervised network
    Histogram equalization of bilateral multi-scale fusion
    Structure diagram of DBSNet
    DBS module
    RAC module
    PD and PD-1 operations
    Output effects of different stride factors and convolution kernel sizes
    Visual comparison of processing results of low-light images by different algorithms. (a) Input image; (b) LIME algorithm[27]; (c) SIRE algorithm[28]; (d) NPE algorithm[29]; (e) EG algorithm[26]; (f) Zero-DCE algorithm[25]; (g) RetinexNet algorithm[6]; (h) DUPE algorithm[32]; (i) LLFLow algorithm[30]; (j) GLADNet algorithm[7]; (k) SCI algorithm[31]; (l) proposed algorithm
    Visual comparison of processing results of real low-light images by different algorithms. (a) Input image; (b) LIME algorithm[27]; (c) SIRE algorithm[28]; (d) RetinexNet algorithm[6]; (e) LLFLow algorithm[30]; (f) EG algorithm[26]; (i) Zero-DCE algorithm[25]; (j) KinD++ algorithm[8]; (k) proposed algorithm
    Ablation experiments with different loss functions. (a) LDBSNet; (b) Lspa; (c) Lcol; (d) Ltv; (e) DBSNet; (f) real scene
    • Table 1. Comparison of output images under different stride factors

      View table

      Table 1. Comparison of output images under different stride factors

      hPSNR↑SSIM↑
      115.600.685
      219.550.723
      320.490.775
      419.210.801
      522.120.836
      620.560.831
      720.320.799
    • Table 2. Comparison of output image indexes under different algorithms

      View table

      Table 2. Comparison of output image indexes under different algorithms

      IndexLIME27SIRE28NPE29EG26Zero-DCE25RetinexNet6
      PSNR↑16.7611.8616.9717.4814.8615.17
      SSIM↑0.4440.4940.4820.6540.5620.075
      NIQE↓9.7798.0739.7885.2388.8116.261
      Delta E↓21.4332.6221.7719.3124.5624.17
      IndexDUPE32LLFlow30GLADNet7SCI31KinD++8Proposed
      PSNR↑14.7719.1919.7219.8120.1220.81
      SSIM↑0.4700.7140.6850.7310.7360.739
      NIQE↓9.0794.2137.2834.1034.2904.191
      Delta E↓26.1916.1716.5415.5911.5211.50
    • Table 3. Ablation experiments with different loss functions

      View table

      Table 3. Ablation experiments with different loss functions

      LDBSNetLspaLcolLtvPSNR↑SSIM↑
      ×10.110.456
      ×18.600.684
      ×20.550.711
      ×17.410.759
      22.210.863
    Tools

    Get Citation

    Copy Citation Text

    Yong Chen, Jinliang Zhang, Huanlin Liu, Kaixin Shao, Shangming Chen, Hangying Xiong, Yourui Zhang. Self-Supervised Enhancement of Low-Light Images Based on Blind Spot Networks[J]. Acta Optica Sinica, 2024, 44(13): 1310001

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Image Processing

    Received: Jan. 23, 2024

    Accepted: Mar. 15, 2024

    Published Online: Jul. 4, 2024

    The Author Email: Yong Chen (chenyong@cqupt.edu.cn)

    DOI:10.3788/AOS240549

    CSTR:32393.14.AOS240549

    Topics