Optics and Precision Engineering, Volume. 32, Issue 13, 2112(2024)

Underwater image enhancement using joint texture perception and color histogram features

Guoming YUAN*... Haijun LIU, Xiaoli LI, Ruilei ZHANG and Weifeng SHAN |Show fewer author(s)
Author Affiliations
  • Department of emergency management, Institute of Disaster Prevention, Sanhe065201, China
  • show less
    Figures & Tables(22)
    Architecture of underwater image enhancement using joint texture perception and color histogram features
    Visual results of V3, C3 and E3 features
    Transformer module
    Multiple-head self attention mechanism
    Architecture of self attention in multi-head attention mechanism of deformable transformer block(where M×M represents the length×width, C is the channel number, reshape denotes the reshape layer)
    Architecture of space aware deformable convolution
    Architecture of texture-aware network
    Architecture of histogram color feature extraction network
    Visualized results of color features with different dimensions
    Predicted histogram results
    Real histogram results
    Architecture of color texture fusion module
    Architecture of color-texture fusion-based underwater image enhancement network
    Visualized results of different variant models
    Comparison of enhanced results on synthetic underwater images
    Comparison of enhanced results on real underwater images
    Comparison of enhanced results on challenging underwater images dataset
    • Table 1. 10种水类型下不同光波长的值

      View table
      View in Article

      Table 1. 10种水类型下不同光波长的值

      TypeRGBTypeRGB
      I0.9820.9610.80510.8750.8850.750
      IA0.9750.9550.80430.8000.8200.710
      IB0.9680.9500.83050.6700.7300.670
      II0.9400.9250.80070.5000.6100.620
      III0.8900.8850.75080.2900.4600.550
    • Table 2. Results of ablation studies on synthetic images

      View table
      View in Article

      Table 2. Results of ablation studies on synthetic images

      ModelPSNRSSIM
      w/o G22.150.853 5
      w/o C22.230.862 2
      Depth-MSA24.110.899 1
      Texture-RB23.740.882 9
      w/o F23.650.880 7
      Ours25.350.903 4
    • Table 3. Comparison of enhanced results by different algorithms on synthetic and real image datasets

      View table
      View in Article

      Table 3. Comparison of enhanced results by different algorithms on synthetic and real image datasets

      ModelSynthetic imagesReal images
      PSNRSSIMEIPCQIUIQMEntropyPSNRSSIMEIPCQIUIQMEntropy
      HUIE411.480.63566.520.9682.3957.02918.510.803 074.890.9862.8277.338
      WDN915.880.72571.970.9712.5567.12315.230.676 368.450.9632.4547.022
      ALAN1015.520.731 172.190.9702.5917.19419.250.871 873.890.9772.8407.417
      Swin-conv1115.480.737 774.550.9732.7517.24123.370.890 782.580.9942.9927.553
      U-T1214.780.575 381.521.0543.1117.82420.670.880 681.451.0372.9867.527
      Ours25.280.903 483.671.1053.2767.95224.630.900 683.551.0963.1057.658
    • Table 4. Comparison of different methods on challenge and UCCS datasets

      View table
      View in Article

      Table 4. Comparison of different methods on challenge and UCCS datasets

      ModelChallenge
      UIQMEntropy
      HUIE2.3415.229
      WDN2.4546.110
      ALAN2.5986.654
      Swin-conv2.6987.012
      U-T2.5776.741
      Ours2.8977.212
    • Table 5. Running time comparison of different algorithms

      View table
      View in Article

      Table 5. Running time comparison of different algorithms

      ModelTime
      HUIE2.351
      WDN0.032
      ALAN0.056
      Swin-conv0.173
      U-T0.183
      Ours0.051
    Tools

    Get Citation

    Copy Citation Text

    Guoming YUAN, Haijun LIU, Xiaoli LI, Ruilei ZHANG, Weifeng SHAN. Underwater image enhancement using joint texture perception and color histogram features[J]. Optics and Precision Engineering, 2024, 32(13): 2112

    Download Citation

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

    Category:

    Received: Nov. 1, 2023

    Accepted: --

    Published Online: Aug. 28, 2024

    The Author Email: YUAN Guoming (scdyuan@126.com)

    DOI:10.37188/OPE.20243213.2112

    Topics