Laser & Optoelectronics Progress, Volume. 61, Issue 15, 1501001(2024)

Reconstruction Method for Measured Atmospheric Polarization Mode Based on Non-Neighborhood Constraint

Wenli Wang1,2, Xinjian Gao1、*, Changhao Bao1,2, Lei Shi1,2, Xin Wang1,2, and Jun Gao1,2
Author Affiliations
  • 1School of Computer and Information, Hefei University of Technology, Hefei 230009, Anhui, China
  • 2Image Information Processing Laboratory, Hefei University of Technology, Hefei 230009, Anhui, China
  • show less
    Figures & Tables(14)
    Acquisition process of ground truth of disturbed atmospheric polarization data
    Reconstruction network of measured atmospheric polarization mode based on non-neighborhood constraints
    Non-neighborhood feature reconstruction process. (a) Data to be reconstructed; (b) reconstruction process; (c) reconstruction results
    Polarization data acquisition equipment
    Polarization reconstruction results under irregular mask coverage. (a) Data to be reconstructed; (b) ground truth; (c) reconstruction result
    Reconstruction results under interference of different areas
    Reconstruction experiment of measured atmospheric polarization mode under the surrounding interference. (a) Sky polarization image; (b) polarization image need to be reconstructed; (c) ground truth; (d) reconstruction result
    Reconstruction experiment of measured atmospheric polarization mode under the surrounding interference. (a) Sky polarization image; (b) polarization image need to be reconstructed; (c) ground truth; (d) reconstruction result
    Polarization reconstruction experiment under cloud interference. (a) Sky polarization image; (b) polarization image need to be reconstructed; (c) ground truth; (d) reconstruction result
    Comparison of reconstruction results of different methods. (a) Polarization image need to be reconstructed; (b) ground truth; (c) Ref. [11] method; (d) Ref. [15] method; (e) Ref.[17] method; (f) MISF[24]; (g) Repaint[25]; (h) proposed method
    Reconstruction results under cloud interference after removing non-neighborhood feature inpainting module. (a) Change trend of the indicator PSNR; (b) change trend of the indicator SSIM
    • Table 1. Quantitative analysis of reconstruction results under cloud disturbance

      View table

      Table 1. Quantitative analysis of reconstruction results under cloud disturbance

      Cloud areaMSEPSNRSSIM
      (0%,20%)0.021427.250.9470
      (20%,40%)0.027626.570.9335
      (40%,60%)0.045424.220.8927
      (60%,80%)0.056222.840.8652
      (80%,100%)0.057722.250.8604
    • Table 2. Quantitative analysis of reconstruction results with different methods

      View table

      Table 2. Quantitative analysis of reconstruction results with different methods

      MethodMSEPSNRSSIM
      Ref.[110.045820.320.7931
      Ref.[150.040221.450.8148
      Ref.[170.041722.260.8226
      MISF240.038123.680.8624
      Repaint250.038323.640.8861
      Proposed0.034624.370.8995
    • Table 3. Reconstruction results under cloud interference after removing non-neighborhood feature inpainting module

      View table

      Table 3. Reconstruction results under cloud interference after removing non-neighborhood feature inpainting module

      Cloud areaMSEPSNRSSIM
      (0%,20%)0.035325.640.8744
      (20%,40%)0.056424.150.8082
      (40%,60%)0.082722.260.7159
      (60%,80%)0.091219.410.6914
      (80%,100%)0.122117.040.5657
    Tools

    Get Citation

    Copy Citation Text

    Wenli Wang, Xinjian Gao, Changhao Bao, Lei Shi, Xin Wang, Jun Gao. Reconstruction Method for Measured Atmospheric Polarization Mode Based on Non-Neighborhood Constraint[J]. Laser & Optoelectronics Progress, 2024, 61(15): 1501001

    Download Citation

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

    Category: Atmospheric Optics and Oceanic Optics

    Received: May. 24, 2023

    Accepted: Aug. 8, 2023

    Published Online: Aug. 12, 2024

    The Author Email: Xinjian Gao (gaoxinjian@hfut.edu.cn)

    DOI:10.3788/LOP231369

    CSTR:32186.14.LOP231369

    Topics