• Laser & Optoelectronics Progress
  • Vol. 60, Issue 2, 0210011 (2023)
Ruihu Cao1, Pengchao Zhang1,2,*, Lei Wang1, Fan Zhang1, and Jie Kang1
Author Affiliations
  • 1School of Mechanical Engineering, Shaanxi University of Technology, Hanzhong 723000, Shaanxi , China
  • 2Shaanxi Key Laboratory of Industrial Automation, Hanzhong 723000, Shaanxi , China
  • show less
    DOI: 10.3788/LOP213235 Cite this Article Set citation alerts
    Ruihu Cao, Pengchao Zhang, Lei Wang, Fan Zhang, Jie Kang. Single Image Defogging Algorithm Based on Attention Mechanism[J]. Laser & Optoelectronics Progress, 2023, 60(2): 0210011 Copy Citation Text show less
    DRS-Dehaze Net framework
    Fig. 1. DRS-Dehaze Net framework
    DRS-Block structure
    Fig. 2. DRS-Block structure
    SKattention structure
    Fig. 3. SKattention structure
    Comparison of dehazing results for synthetic foggy images. (a) Foggy images; (b) original images; (c) AOD; (d) BCCR; (e) DCP; (f) Dehazenet; (g) Yoly; (h) proposed algorithm
    Fig. 4. Comparison of dehazing results for synthetic foggy images. (a) Foggy images; (b) original images; (c) AOD; (d) BCCR; (e) DCP; (f) Dehazenet; (g) Yoly; (h) proposed algorithm
    Result comparison for real fogged images Girls. (a) Foggy image; (b) AOD; (c) BCCR; (d) DCP; (e) Dehazenet; (f) Yoly; (g) proposed algorithm
    Fig. 5. Result comparison for real fogged images Girls. (a) Foggy image; (b) AOD; (c) BCCR; (d) DCP; (e) Dehazenet; (f) Yoly; (g) proposed algorithm
    Result comparison for real fogged images Trees. (a) Foggy image; (b) AOD; (c) BCCR; (d) DCP; (e) Dehazenet; (f) Yoly; (g) proposed algorithm
    Fig. 6. Result comparison for real fogged images Trees. (a) Foggy image; (b) AOD; (c) BCCR; (d) DCP; (e) Dehazenet; (f) Yoly; (g) proposed algorithm
    Result comparison for real fogged images Tian An Men. (a) Foggy image; (b) AOD; (c) BCCR; (d) DCP; (e) Dehazenet; (f) Yoly; (g) proposed algorithm
    Fig. 7. Result comparison for real fogged images Tian An Men. (a) Foggy image; (b) AOD; (c) BCCR; (d) DCP; (e) Dehazenet; (f) Yoly; (g) proposed algorithm
    Comparison of average gradient for real fogged images
    Fig. 8. Comparison of average gradient for real fogged images
    Comparison of information entropy for real fogged images
    Fig. 9. Comparison of information entropy for real fogged images
    AlgorithmEvaluation indicatorConference roomRed curtainGreen tabletBedroomSitting roomAverage
    AODPSNR16.01110.54414.73415.44415.54714.456
    SSIM0.8210.6160.7970.7280.7680.746
    BCCRPSNR11.8705.07514.36516.3099.08111.340
    SSIM0.4700.2540.6780.4640.5040.474
    DCPPSNR17.24913.72119.14914.88916.89716.381
    SSIM0.7140.6340.8210.7970.6530.724
    DehazenetPSNR18.66010.51016.61617.29616.22815.862
    SSIM0.8900.6280.8640.7430.7980.785
    YolyPSNR16.46015.16116.34116.20416.33316.100
    SSIM0.6490.6760.7260.7420.5760.674
    Proposed algorithmPSNR23.36711.88819.71517.80819.55318.466
    SSIM0.8690.6250.8720.7760.8650.801
    Table 1. Experimental results for synthetic fogged images
    ParameterDehazenetDCPMCF-DehazenetGriddehazenetProposed algorithm
    Time /s1.839.823.051.951.92
    Table 2. Comparison of running time of different algorithms
    NetworkGirlsTreesTian An Men
    DR4.5259.3093.382
    DRS5.84111.3993.789
    Table 3. Average gradient of ablation experiments
    Ruihu Cao, Pengchao Zhang, Lei Wang, Fan Zhang, Jie Kang. Single Image Defogging Algorithm Based on Attention Mechanism[J]. Laser & Optoelectronics Progress, 2023, 60(2): 0210011
    Download Citation