Acta Photonica Sinica, Volume. 49, Issue 7, 709001(2020)

Nonlinear Reconstruction for Off-axis Fresnel Digital Holography with Deep Learning

Hang LIU... Yong-liang XIAO*, Jun-long TIAN, Hong-xing LI and Jian-xin ZHONG |Show fewer author(s)
Author Affiliations
  • School of Physics and Optoelectronic Engineering, Xiangtan University, Xiangtan, Hunan 411105, China
  • show less
    Figures & Tables(13)
    Schematic diagram of off⁃axis Fresnel digital hologram recording
    Method for off⁃axis Fresnel digital hologram reconstruction based on deep learning
    The structure model of ResNet
    Structure of each module in ResNet model
    Training samples of numerical simulation experiment
    The training loss curves in ResNet training by using L2 loss functions
    Schematic diagram of off⁃axis Fresnel digital hologram recording
    Off⁃axis Fresnel digital hologram nonlinear reconstruction with ResNet
    ResNet reconstruction results of test dateset with different diffraction distances (plane wave training, z0=0.3 m)
    Distance Robust ResNet reconstruction results for different diffraction distance (plane wave)
    The MAE, RMSE, and SSIM of Distance Robust ResNet reconstruction results and corresponding object images for different diffraction distance (plane wave)
    • Table 1. Compare MAE, RMSE and SSIM for deep learning object reconstruction with traditional algorithms

      View table
      View in Article

      Table 1. Compare MAE, RMSE and SSIM for deep learning object reconstruction with traditional algorithms

      Frequency filteringFour⁃step phase shiftPlane z0=0.3 mPlane z0=0.4 mPlane z0=0.5 mSpherical z0=0.3 mSpherical z0=0.4 mSpherical z0=0.5 m
      MAE1.870 51.808 3↓1.212 01.398 01.524 71.932 52.111 22.200 9
      RMSE2.677 22.515 0↓2.208 72.295 02.427 32.609 02.688 52.675 2
      SSIM0.729 10.740 8↓0.918 00.909 10.906 60.888 60.883 10.879 6
    • Table 2. Compare MAE, RMSE and SSIM for distance⁃robust with single diffraction distance ResNet reconstruction results

      View table
      View in Article

      Table 2. Compare MAE, RMSE and SSIM for distance⁃robust with single diffraction distance ResNet reconstruction results

      Plane z0=0.3 mPlane z0=0.4 mPlane z0=0.5 mRobust z0=0.3 mRobust z0=0.4 mRobust z0=0.5 m
      MAE↓1.212 01.398 01.524 72.870 81.744 91.552 4
      RMSE↓2.208 72.295 02.427 33.037 12.599 32.463 5
      SSIM↑0.918 00.909 10.906 60.889 00. 891 90.882 9
    Tools

    Get Citation

    Copy Citation Text

    Hang LIU, Yong-liang XIAO, Jun-long TIAN, Hong-xing LI, Jian-xin ZHONG. Nonlinear Reconstruction for Off-axis Fresnel Digital Holography with Deep Learning[J]. Acta Photonica Sinica, 2020, 49(7): 709001

    Download Citation

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

    Category: Holography

    Received: Mar. 26, 2020

    Accepted: --

    Published Online: Aug. 25, 2020

    The Author Email: XIAO Yong-liang (ylxiao@xtu.edu.cn)

    DOI:10.3788/gzxb20204907.0709001

    Topics