• Laser & Optoelectronics Progress
  • Vol. 58, Issue 20, 2010012 (2021)
Yanfei Peng**, Pingjia Zhang*, Yi Gao, and Lingling Zi
Author Affiliations
  • School of Electronic and Information Engineering, Liaoning Technical University, Huludao, Liaoning 125105, China
  • show less
    DOI: 10.3788/LOP202158.2010012 Cite this Article Set citation alerts
    Yanfei Peng, Pingjia Zhang, Yi Gao, Lingling Zi. Attention Fusion Generative Adversarial Network for Single-Image Super-Resolution Reconstruction[J]. Laser & Optoelectronics Progress, 2021, 58(20): 2010012 Copy Citation Text show less
    Generator network structure
    Fig. 1. Generator network structure
    Discriminator network structure
    Fig. 2. Discriminator network structure
    Use of channel and spatial attention modules
    Fig. 3. Use of channel and spatial attention modules
    Reconstruction effects with different values of ε coefficient
    Fig. 4. Reconstruction effects with different values of ε coefficient
    Comparison of residual blocks. (a)SRGAN; (b) proposed model
    Fig. 5. Comparison of residual blocks. (a)SRGAN; (b) proposed model
    Variation curve of generator function loss value
    Fig. 6. Variation curve of generator function loss value
    Variation curve of discriminant function loss value
    Fig. 7. Variation curve of discriminant function loss value
    Partial enlarged comparison diagrams of the “baby” reconstruction effect of five algorithms in Set5 test set
    Fig. 8. Partial enlarged comparison diagrams of the “baby” reconstruction effect of five algorithms in Set5 test set
    Partial enlarged comparison diagrams of the “butterfly” reconstruction effect of five algorithms in Set5 test set
    Fig. 9. Partial enlarged comparison diagrams of the “butterfly” reconstruction effect of five algorithms in Set5 test set
    Partial enlarged comparison diagrams of the “pepper” reconstruction effect of five algorithms in Set14 test set
    Fig. 10. Partial enlarged comparison diagrams of the “pepper” reconstruction effect of five algorithms in Set14 test set
    Partial enlarged comparison diagrams of the “fish” reconstruction effect of five algorithms in BSDS100 test set
    Fig. 11. Partial enlarged comparison diagrams of the “fish” reconstruction effect of five algorithms in BSDS100 test set
    Partial enlarged comparison diagrams of the “room” reconstruction effect of five algorithms in Urban100 test set
    Fig. 12. Partial enlarged comparison diagrams of the “room” reconstruction effect of five algorithms in Urban100 test set
    Partial enlarged comparison diagrams of the “baby” reconstruction effect in ablation experiment in Set5 test set
    Fig. 13. Partial enlarged comparison diagrams of the “baby” reconstruction effect in ablation experiment in Set5 test set
    Partial enlarged comparison diagrams of the “butterfly” reconstruction effect in ablation experiment in Set5 test set
    Fig. 14. Partial enlarged comparison diagrams of the “butterfly” reconstruction effect in ablation experiment in Set5 test set
    Partial enlarged comparison diagrams of the “lenna” reconstruction effect in ablation experiment in Set14 test set
    Fig. 15. Partial enlarged comparison diagrams of the “lenna” reconstruction effect in ablation experiment in Set14 test set
    Test setScaleBicubicESPCNSRGANESRGANProposed
    Set526.69227.59426.62828.54329.510
    Set1424.56525.18624.56824.53226.443
    Urban10021.70622.30022.11322.79223.917
    BSDS10024.64125.04324.47225.32225.884
    Table 1. Comparison of PSNR values of various super-resolution reconstruction methods
    Test setScaleBicubicESPCNSRGANESRGANProposed
    Set50.77300.78950.80230.81450.8517
    Set140.67320.69830.70190.67110.7377
    Urban1000.63170.65950.67740.70500.7415
    BSDS1000.64010.67020.67130.65140.7002
    Table 2. Comparison of SSIM values of various super-resolution reconstruction methods
    MethodSet5Set14Urban100BSDS100
    SRGAN26.62824.56822.11324.472
    SRGAN-BN27.86325.26922.74325.228
    SRGAN+CA&SA27.86525.37522.69325.198
    SRGAN+Charbonnier28.15325.72422.91925.404
    SRGAN+Charbonnier+CA&SA-BN29.51026.44323.91725.884
    Table 3. PSNR values of models with different module combinations on four test sets
    MethodSet5Set14UrbanBSDS100
    SRGAN0.80230.70190.67740.6713
    SRGAN-BN0.80580.70790.68200.6728
    SRGAN+CA&SA0.80570.70680.67920.6754
    SRGAN+Charbonnier0.82030.71800.69390.6822
    SRGAN+Charbonnier+A&SA-BN0.85170.73770.74150.7002
    Table 4. SSIM values of models with different module combinations on four test sets
    Yanfei Peng, Pingjia Zhang, Yi Gao, Lingling Zi. Attention Fusion Generative Adversarial Network for Single-Image Super-Resolution Reconstruction[J]. Laser & Optoelectronics Progress, 2021, 58(20): 2010012
    Download Citation