Optoelectronic Technology, Volume. 41, Issue 3, 185(2021)

Computational Optical Imaging through Scattering Media by Generative Adversarial Networks

Xiren ZHANG, Hengjing ZHANG, Yaling LUO, and Lifeng YANG
Author Affiliations
  • School of Optoelectronic Science and Engineering, University of Electronic Science and Technology of China, Chengdu 610000, CHN
  • show less
    Figures & Tables(7)
    Optical path diagram of optical scattering imaging system
    Structural diagram of Generative Adversarial Network
    The loss function curves of discriminator and generator when L1 norm is used as content loss function under ground glass scattering
    Comparison of the results generated by the network models of the two scattering media at 1,10,100,500,1000 periods
    Comparison of reconstructed graphs of two data sets under three mixed loss functions
    • Table 1. Evaluation index values of reconstructed images from MINIST data sets

      View table
      View in Article

      Table 1. Evaluation index values of reconstructed images from MINIST data sets

      损失函数SSIMPSNRPCC
      GAN+L1_loss0.71616.7870.850
      GAN+L2_loss0.70716.4780.845
      GAN+SSIM_loss0.75618.1490.896
    • Table 2. Evaluation index values of reconstructed images from Fashion⁃MINIST data sets

      View table
      View in Article

      Table 2. Evaluation index values of reconstructed images from Fashion⁃MINIST data sets

      损失函数SSIMPSNRPCC
      GAN+L1_loss0.53514.1110.851
      GAN+L2_loss0.53413.9770.840
      GAN+SSIM_loss0.55714.2700.887
    Tools

    Get Citation

    Copy Citation Text

    Xiren ZHANG, Hengjing ZHANG, Yaling LUO, Lifeng YANG. Computational Optical Imaging through Scattering Media by Generative Adversarial Networks[J]. Optoelectronic Technology, 2021, 41(3): 185

    Download Citation

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

    Category: Research and Trial-manufacture

    Received: Feb. 28, 2021

    Accepted: --

    Published Online: Oct. 26, 2021

    The Author Email:

    DOI:10.19453/j.cnki.1005-488x.2021.03.005

    Topics