Laser & Optoelectronics Progress, Volume. 56, Issue 11, 111005(2019)

Single-Image Defogging Algorithm Based on Deep Learning

Jiantang Zhao*
Author Affiliations
  • College of Mathematics and Information Science, Xianyang Normal University, Xianyang, Shaanxi 712000, China
  • show less
    Figures & Tables(14)
    Schematic of atmospheric scattering model
    Proposed network structure
    Flow chart of proposed algorithm
    Defogging results of foggy image Teddy by different algorithms. (a) Foggy image; (b) original clear image; (c) DCP algorithm; (d) BCCR algorithm; (e) SVDSR algorithm; (f) CAP algorithm; (g) MSCNN algorithm; (h) proposed algorithm
    Defogging results of foggy image Dolls by different algorithms. (a) Foggy image; (b) original clear image; (c) DCP algorithm; (d) BCCR algorithm; (e) SVDSR algorithm; (f) CAP algorithm; (g) MSCNN algorithm; (h) proposed algorithm
    Defogging results of foggy image Cloth by different algorithms. (a) Foggy image; (b) original clear image; (c) DCP algorithm; (d) BCCR algorithm; (e) SVDSR algorithm; (f) CAP algorithm; (g) MSCNN algorithm; (h) proposed algorithm
    Comparison of defogging results of natural foggy image House. (a) Foggy image; (b) DCP algorithm; (c) BCCR algorithm; (d) SVDSR algorithm; (e) CAP algorithm; (f) MSCNN algorithm; (g) proposed algorithm
    Comparison of defogging results of natural foggy image Pumpkin. (a) Foggy image; (b) DCP algorithm; (c) BCCR algorithm; (d) SVDSR algorithm; (e) CAP algorithm; (f) MSCNN algorithm; (g) proposed algorithm
    Comparison of defogging results of natural foggy image Girls. (a) Foggy image; (b) DCP algorithm; (c) BCCR algorithm; (d) SVDSR algorithm; (e) CAP algorithm; (f) MSCNN algorithm; (g) proposed algorithm
    Comparison results of different algorithms. (a) Average gradient; (b) information entropy
    • Table 1. Multi-scale convolution parameters

      View table

      Table 1. Multi-scale convolution parameters

      Filter sizePadStride
      1×1×1601
      3×3×165×5×167×7×16123111
    • Table 2. Evaluation indicators of defogging results of image Teddy by different algorithms

      View table

      Table 2. Evaluation indicators of defogging results of image Teddy by different algorithms

      IndicatorDCPBCCRSVDSRCAPMSCNNProposed
      RMSE ↓UQI ↑Cross entropy ↑Tone reduction↑Average gradient ↑Entropy ↑PSNR /dB ↑0.02730.61460.57630.756311.004417.026515.82460.01330.56781.13160.24328.806913.351312.59620.06290.62450.36430.75019.831814.628015.25100.02590.61791.63660.79087.149916.799519.87020.02580.60351.20880.69667.979916.385520.60220.02460.63271.25290.844111.195617.883123.4425
      SSIM ↑0.77820.60970.75720.87690.87970.9524
    • Table 3. Evaluation indicators of defogging results of image Dolls by different algorithms

      View table

      Table 3. Evaluation indicators of defogging results of image Dolls by different algorithms

      IndicatorDCPBCCRSVDSRCAPMSCNNProposed
      RMSE ↓UQI ↑Cross entropy ↑Tone reduction↑Average gradient ↑Entropy ↑PSNR /dB↑0.03200.59470.23000.93046.274614.980111.48450.02270.64402.54090.34556.956013.321810.65210.07560.67240.25360.76947.469613.741719.49850.03130.61591.37980.52363.949414.551824.65580.02970.59550.68060.55774.367114.313222.32590.02990.67812.58200.98437.556216.790824.7741
      SSIM ↑0.84120.63390.86010.87690.85830.9245
    • Table 4. Evaluation indicators of defogging results of image Cloth by different algorithms

      View table

      Table 4. Evaluation indicators of defogging results of image Cloth by different algorithms

      IndicatorDCPBCCRSVDSRCAPMSCNNProposed
      RMSE ↓UQI ↑Cross entropy ↑Tone reduction↑Average gradient ↑Entropy ↑PSNR /dB↑0.03750.83781.01920.745916.562913.314924.23850.02870.92201.43420.624322.515215.109816.26980.09660.88381.02650.63237.326912.659815.25020.02310.52310.22600.80455.614615.540523.49580.02410.68950.48520.65565.231214.231921.21020.02250.98670.04210.965022.768016.699527.3441
      SSIM ↑0.85670.73570.72790.94620.89750.9690
    Tools

    Get Citation

    Copy Citation Text

    Jiantang Zhao. Single-Image Defogging Algorithm Based on Deep Learning[J]. Laser & Optoelectronics Progress, 2019, 56(11): 111005

    Download Citation

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

    Category: Image Processing

    Received: Dec. 6, 2018

    Accepted: Dec. 25, 2018

    Published Online: Jun. 13, 2019

    The Author Email: Zhao Jiantang (zhaojiantanglaoshi@126.com)

    DOI:10.3788/LOP56.111005

    Topics