Photonics Research, Volume. 9, Issue 5, B210(2021)

Imaging through unknown scattering media based on physics-informed learning

Shuo Zhu1、†, Enlai Guo1,2、†,*, Jie Gu1, Lianfa Bai1, and Jing Han1,3、*
Author Affiliations
  • 1Jiangsu Key Laboratory of Spectral Imaging and Intelligent Sense, Nanjing University of Science and Technology, Nanjing 210094, China
  • 2e-mail: njustgel@njust.edu.cn
  • 3e-mail: eohj@njust.edu.cn
  • show less
    Figures & Tables(15)
    Speckle statistical characteristics analysis of the same object corresponding to different testing diffusers. (a) First row and second row are the speckle autocorrelation of the object within or exceeding the OME range, the third row is the cross-correlation with D1, respectively. (b)–(d) Intensity values of the white dash lines in the first, second, and third rows of (a), respectively. The color bar represents the normalized intensity. Scale bars: 875.52 µm.
    Schematic of the physics-informed learning method for scalable scattering imaging.
    Experimental setup for the scalable imaging. Different diffusers are employed to obtain speckle patterns with different scattering scenes. The OME range of this system is also measured by calculating the cross-correlation coefficient [21]. See Appendix B for details.
    Testing results for generalization reconstruction of Group 1. Scale bars: 264.24 µm.
    Testing results for generalization reconstruction of Group 2. Scale bars: 264.24 µm.
    Testing results for generalization reconstruction of Group 3. Scale bars: 264.24 µm.
    Testing results for generalization reconstruction of Group 4. Scale bars: 820.8 µm.
    Generalization results for a single-character object with different scales and the scale of FOV is defined as the FOV/OME times. (a), (b) Results with different amounts of training diffusers, which are trained with one diffuser and three diffusers, respectively. (c) Reconstruction results with different scales and corresponding ground truth (GT).
    Comparison results without or with this pre-processing step for imaging through an unknown diffuser. Three ground glasses are selected as the training diffusers and another diffuser for testing.
    Results with different number of speckles via the physics-informed learning method. Three ground glasses are selected as the training diffusers and another diffuser for testing.
    Generalization results of imaging exceeding OME range with different complexity objects.
    • Table 1. Quantitative Evaluation Results of the Objects within OME

      View table
      View in Article

      Table 1. Quantitative Evaluation Results of the Objects within OME

      TypeTraining SetTesting SetMAESSIMPSNR (dB)
      Group 1D1Seen objects0.01900.847723.41
      D1Unseen objects0.02880.824819.86
      D1, D2, D3Seen objects0.00330.974440.56
      D1, D2, D3Unseen objects0.02710.808820.58
      Group 2D1Seen objects0.02570.860820.75
      D1Unseen objects0.05480.692916.56
      D1, D2, D3Seen objects0.00720.919038.60
      D1, D2, D3Unseen objects0.04550.767816.76
      Group 3D1Seen objects0.11660.467215.59
      D1Unseen objects0.12340.535114.98
      D1, D2, D3Seen objects0.03230.852323.82
      D1, D2, D3Unseen objects0.05340.741020.67
    • Table 2. Quantitative Evaluation Results of Objects Extending the FOV 1.2 Times

      View table
      View in Article

      Table 2. Quantitative Evaluation Results of Objects Extending the FOV 1.2 Times

      TypeTraining SetTesting SetMAESSIMPSNR (dB)
      Group 4D1Seen objects0.03590.843917.53
      D1Unseen objects0.04320.696817.12
      D1, D2, D3Seen objects0.00850.943130.55
      D1, D2, D3Unseen objects0.03090.870618.90
    • Table 3. Objective Indicators with Different Number of Speckles via the Physics-Informed Learning Method

      View table
      View in Article

      Table 3. Objective Indicators with Different Number of Speckles via the Physics-Informed Learning Method

      Number of SpecklesMAESSIMPSNR (dB)
      256×2560.00540.934238.38
      512×5120.00370.953944.53
    • Table 4. Objective Indicators Corresponding to Fig. 11

      View table
      View in Article

      Table 4. Objective Indicators Corresponding to Fig. 11

      ObjectsNetworkMAESSIMPSNR (dB)
      5× OME single charactersU-Net0.02320.874121.70
      5× OME double charactersU-Net0.03610.836018.96
      1.2× OME facesU-Net0.06730.697218.40
      1.2× OME facesPDSNet-L0.05670.705021.16
    Tools

    Get Citation

    Copy Citation Text

    Shuo Zhu, Enlai Guo, Jie Gu, Lianfa Bai, Jing Han. Imaging through unknown scattering media based on physics-informed learning[J]. Photonics Research, 2021, 9(5): B210

    Download Citation

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

    Special Issue: DEEP LEARNING IN PHOTONICS

    Received: Dec. 2, 2020

    Accepted: Feb. 21, 2021

    Published Online: Apr. 19, 2021

    The Author Email: Enlai Guo (njustgel@njust.edu.cn), Jing Han (eohj@njust.edu.cn)

    DOI:10.1364/PRJ.416551

    Topics