• Laser & Optoelectronics Progress
  • Vol. 59, Issue 2, 0209001 (2022)
Jian Pu, Jinbin Gui*, and Kai Zhang
Author Affiliations
  • Faculty of Science, Kunming University of Science and Technology, Kunming , Yunnan 650550, China
  • show less
    DOI: 10.3788/LOP202259.0209001 Cite this Article Set citation alerts
    Jian Pu, Jinbin Gui, Kai Zhang. Multiscale Digital Hologram Reconstruction Based on Deep Learning[J]. Laser & Optoelectronics Progress, 2022, 59(2): 0209001 Copy Citation Text show less
    Schematic of holographic wavefront recording
    Fig. 1. Schematic of holographic wavefront recording
    Schematic of digital holographic reconstruction
    Fig. 2. Schematic of digital holographic reconstruction
    Improved U-Net structure, the size of the current feature map is located at the left of the rectangular box
    Fig. 3. Improved U-Net structure, the size of the current feature map is located at the left of the rectangular box
    HS-Block structure
    Fig. 4. HS-Block structure
    Network structure for reconstructing multi-scale digital holograms
    Fig. 5. Network structure for reconstructing multi-scale digital holograms
    Mixing digital holograms with different scales as a data set training deep learning model to compare the reconstruction effect of different resolution digital holograms
    Fig. 6. Mixing digital holograms with different scales as a data set training deep learning model to compare the reconstruction effect of different resolution digital holograms
    Comparison of reconstruction results of different scale digital holograms by a single deep learning model
    Fig. 7. Comparison of reconstruction results of different scale digital holograms by a single deep learning model
    Parameter256×256512×512640×480
    AmplitudePhaseAmplitudePhaseAmplitudePhase
    PSNR /dB32.470825.416745.837628.313041.465137.7251
    SSIM0.96890.86850.99810.87480.99530.9131
    Table 1. Average PSNR and SSIM of test set
    Jian Pu, Jinbin Gui, Kai Zhang. Multiscale Digital Hologram Reconstruction Based on Deep Learning[J]. Laser & Optoelectronics Progress, 2022, 59(2): 0209001
    Download Citation