Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 61, Issue 18, 1800003(2024)

Review of Light Field Super-Resolution Algorithm Based on Deep Learning

Yawei Xiong, Anzhi Wang*, and Kaili Zhang
Author Affiliations
  • School of Big Data and Computer Science, Guizhou Normal University, Guiyang 550025, Guizhou, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (13)
    Equations (3)
    References (74)
    Tools
    Paper Information
    Topics
    Figures & Tables(13)
    Biplane light field

    Fig. 1. Biplane light field

    Download full size
    Visualizations of 4D light field

    Fig. 2. Visualizations of 4D light field

    Download full size
    General structure of LFSSR

    Fig. 3. General structure of LFSSR

    Download full size
    General structure of LFASR

    Fig. 4. General structure of LFASR

    Download full size
    General structure of the LFSASR

    Fig. 5. General structure of the LFSASR

    Download full size
    Samples of LF datasets

    Fig. 6. Samples of LF datasets

    Download full size
    Visual results of different LFASR methods

    Fig. 7. Visual results of different LFASR methods

    Download full size
    Visual results of different LFSASR methods

    Fig. 8. Visual results of different LFSASR methods

    Download full size
    Visual results of different LFSSR methods

    Fig. 9. Visual results of different LFSSR methods

    Download full size

    • Table 1. Common used datasets of LFSR

      View table

      Table 1. Common used datasets of LFSR

      Dataset nameCategoryAmount（categories/scenes）Depth map（yes/no）
      HCInew［66］Synthetic24 categoriesYes
      HCIold［67］Synthetic13 scenesYes
      EPFL［68］Real-world10 categoriesYes
      INRIA［69］Real-world109 scenesNo
      STFGantry［70］Real-world22 scenesNo
      STFLytro［71］Real-world9 categoriesYes

    • Table 2. Performance comparison (PSNR/SSIM) of different LFASR methods on different datasets

      View table

      Table 2. Performance comparison (PSNR/SSIM) of different LFASR methods on different datasets

      MethodDataset
      HCInewHCIoldINRIA
      VSLFC［74］32.85/0.90938.58/0.94432.53/0.899
      CNN-EPI［45］26.64/0.74431.43/0.85025.05/0.740
      ShearedEPI［40］31.84/0.89837.61/0.94232.35/0.911
      MALFRNet［49］33.46/0.91640.36/0.94933.79/0.912

    • Table 3. Performance comparison (PSNR/SSIM) of different LFSASR methods at ×2 and ×4 magnifications

      View table

      Table 3. Performance comparison (PSNR/SSIM) of different LFSASR methods at ×2 and ×4 magnifications

      MethodScaleParameterDataset
      EPFLHCInewHCIoldINRIASTFgantry
      Bicubic×2–29.74/0.93831.89/0.93637.69/0.97931.33/0.95831.06/0.950
      LF-InterNet［60］×25.04M34.14/0.97737.28/0.97644.45/0.99535.80/0.98438.72/0.991
      DPT［64］×23.73M34.48/0.97637.35/0.97744.31/0.99436.40/0.98439.52/0.993
      LFT［63］×21.11M34.80/0.97937.84/0.97944.52/0.99536.59/0.98640.51/0.994
      EPIT［65］×21.42M34.83/0.97838.23/0.98145.08/0.99536.67/0.98542.17/0.996
      Bicubic×4–25.14/0.83227.61/0.85232.42/0.93426.82/0.88725.93/0.845
      LF-InterNet［60］×45.48M28.67/0.91630.98/0.91637.11/0.97230.64/0.94930.53/0.941
      DPT［64］×43.78M28.93/0.91731.19/0.91937.39/0.97230.96/0.95031.14/0.949
      LFT［63］×41.16M29.25/0.92131.46/0.92237.63/0.97431.20/0.95231.86/0.955
      EPIT［65］×41.47M29.34/0.92031.51/0.92337.68/0.97431.27/0.95332.18/0.957

    • Table 4. Performance comparison (PSNR/SSIM) of different LFSSR methods at ×2 and ×4 magnifications

      View table

      Table 4. Performance comparison (PSNR/SSIM) of different LFSSR methods at ×2 and ×4 magnifications

      MethodScaleParameterDataset
      EPFLHCInewHCIoldINRIASTFgantry
      Bicubic×2–29.74/0.93831.89/0.93637.69/0.97931.33/0.95831.06/0.950
      EDSR［28］×238.62 M33.09/0.96334.83/0.95941.01/0.98734.96/0.97636.30/0.982
      resLF［29］×27.982 M33.62/0.97136.69/0.97443.42/0.99335.40/0.98038.35/0.990
      LF-DFnet［30］×23.940 M34.51/0.97637.42/0.97744.20/0.99436.42/0.98439.43/0.993
      DistgSSR［34］×23.532 M34.80/0.97937.95/0.98044.92/0.99536.58/0.98640.37/0.994
      Bicubic×4–25.26/0.83227.72/0.85232.58/0.93426.95/0.88726.09/0.842
      EDSR［28］×438.89 M27.83/0.88529.59/0.88635.18/0.95329.66/0.92528.70/0.907
      resLF［29］×48.646 M28.26/0.90430.72/0.91136.71/0.96830.34/0.94130.19/0.932
      LF-DFnet［30］×43.990 M28.77/0.91731.23/0.92037.32/0.97230.83/0.95031.15/0.944
      DistgSSR［34］×43.582 M28.98/0.91931.38/0.92237.55/0.97330.99/0.95231.63/0.953
    Tools

    Get Citation

    Copy Citation Text

    Yawei Xiong, Anzhi Wang, Kaili Zhang. Review of Light Field Super-Resolution Algorithm Based on Deep Learning[J]. Laser & Optoelectronics Progress, 2024, 61(18): 1800003

    Download Citation

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

    Category: Reviews

    Received: Jan. 16, 2024

    Accepted: Mar. 13, 2024

    Published Online: Sep. 9, 2024

    The Author Email: Anzhi Wang (wanganzhi@gznu.edu.cn)

    DOI:10.3788/LOP240543

    CSTR:32186.14.LOP240543

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology