Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 62, Issue 6, 0637012(2025)

Low-Light Image Enhancement Algorithm with Light Perception Enhancement and Dense Residual Denoising

Boran Yang1、*, Zilong Du1, Yong Wang1, Lijun Jiang1, and Wenming Yang2
Author Affiliations
  • 1School of Liangjiang Artificial Intelligence, Chongqing University of Technology, Chongqing 401135, China
  • 2Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen 518055, Guangdong , China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (11)
    Equations (15)
    References (33)
    Tools
    Paper Information
    Topics
    Figures & Tables(11)
    Overall network architecture

    Fig. 1. Overall network architecture

    Download full size
    Multilevel feature extraction module. (a) Height axis multihead attention feature extraction; (b) width axis multihead attention feature extraction; (c) local feature extraction

    Fig. 2. Multilevel feature extraction module. (a) Height axis multihead attention feature extraction; (b) width axis multihead attention feature extraction; (c) local feature extraction

    Download full size
    Multilevel attention fusion module

    Fig. 3. Multilevel attention fusion module

    Download full size
    Enhanced image denoising module

    Fig. 4. Enhanced image denoising module

    Download full size
    Qualitative analysis results of image 1 on LOL dataset. (a) Input; (b) KinD; (c) KinD++; (d) ZeroDCE; (e) EnlightenGAN; (f) Uformer; (g) Restormer; (h) LLFormer; (i) GLLNet; (j) ground truth

    Fig. 5. Qualitative analysis results of image 1 on LOL dataset. (a) Input; (b) KinD; (c) KinD++; (d) ZeroDCE; (e) EnlightenGAN; (f) Uformer; (g) Restormer; (h) LLFormer; (i) GLLNet; (j) ground truth

    Download full size
    Qualitative analysis results of image 2 on LOL dataset. (a) Input; (b) KinD; (c) KinD++; (d) ZeroDCE; (e) EnlightenGAN; (f) Uformer; (g) Restormer; (h) LLFormer; (i) GLLNet; (j) ground truth

    Fig. 6. Qualitative analysis results of image 2 on LOL dataset. (a) Input; (b) KinD; (c) KinD++; (d) ZeroDCE; (e) EnlightenGAN; (f) Uformer; (g) Restormer; (h) LLFormer; (i) GLLNet; (j) ground truth

    Download full size
    Qualitative analysis results on MIT-Adobe FiveK dataset. (a) Input; (b) RetinexNet; (c) EnlightenGAN; (d) MIRNet; (e) DSLR; (f) Uformer; (g) Restormer; (h) LLFormer; (i) GLLNet; (j) ground truth

    Fig. 7. Qualitative analysis results on MIT-Adobe FiveK dataset. (a) Input; (b) RetinexNet; (c) EnlightenGAN; (d) MIRNet; (e) DSLR; (f) Uformer; (g) Restormer; (h) LLFormer; (i) GLLNet; (j) ground truth

    Download full size

    • Table 1. Quantitative analysis results on LOL dataset

      View table

      Table 1. Quantitative analysis results on LOL dataset

      MethodPSNR /dB↑SSIM ↑LPIPS ↓
      RetinexNet16.770.560.742
      ZeroDCE13.550.540.222
      KinD20.870.800.160
      EnlightenGAN17.480.650.153
      KinD++21.300.800.159
      DLN21.940.810.121
      Uformer18.550.720.322
      Restormer22.370.810.142
      LLFormer22.540.790.111
      GLLNet23.530.820.096
      LLFormer（3000 epoch）23.430.810.094
      GLLNet（3000 epoch）24.290.830.087

    • Table 2. Quantitative analysis results on MIT-Adobe FiveK dataset

      View table

      Table 2. Quantitative analysis results on MIT-Adobe FiveK dataset

      MethodPSNR /dB↑SSIM ↑LPIPS ↓
      RetinexNet12.510.670.25
      ZeroDCE15.530.650.16
      DSLR20.240.830.08
      EnlightenGAN17.910.840.14
      KinD++21.990.800.12
      MIRNet23.730.900.04
      Uformer21.920.870.06
      Restormer24.920.900.06
      LLFormer25.210.900.04
      GLLNet25.460.910.03
      LLFormer （1000 epoch）25.360.900.033
      GLLNet （1000 epoch）25.580.920.030

    • Table 3. Results of ablation experiments on LOL dataset

      View table

      Table 3. Results of ablation experiments on LOL dataset

      MFEMAFEIDLOL
      HAEWAELFEPSNR /dB↑SSIM↑LPIPS↓
      √√√√23.150.800.098
      √√√√23.070.820.087
      √√√√23.780.820.087
      √√√√23.250.800.097
      √√√√23.720.810.089
      √√√√√24.290.830.087

    • Table 4. Results of ablation experiments on MIT-Adobe FiveK dataset

      View table

      Table 4. Results of ablation experiments on MIT-Adobe FiveK dataset

      MFEMAFEIDMIT-Adobe FiveK
      HAEWAELFEPSNR /dB↑SSIM↑LPIPS↓
      √√√√25.470.890.032
      √√√√25.510.900.031
      √√√√25.480.900.031
      √√√√25.310.890.031
      √√√√25.430.900.030
      √√√√√25.580.920.030
    Tools

    Get Citation

    Copy Citation Text

    Boran Yang, Zilong Du, Yong Wang, Lijun Jiang, Wenming Yang. Low-Light Image Enhancement Algorithm with Light Perception Enhancement and Dense Residual Denoising[J]. Laser & Optoelectronics Progress, 2025, 62(6): 0637012

    Download Citation

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

    Category: Digital Image Processing

    Received: Aug. 13, 2024

    Accepted: Sep. 10, 2024

    Published Online: Mar. 13, 2025

    The Author Email:

    DOI:10.3788/LOP241837

    CSTR:32186.14.LOP241837

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology