• Acta Photonica Sinica
  • Vol. 51, Issue 3, 0310006 (2022)
Jinxin XU1、2, Qingwu LI2、*, Zhiqiang GUAN1, and Xiaolin WANG2
Author Affiliations
  • 1Nanjing Marine Radar Institute,Nanjing 211106,China
  • 2College of Internet of Things Engineering,Hohai University,Changzhou ,Jiangsu 213002,China
  • show less
    DOI: 10.3788/gzxb20225103.0310006 Cite this Article
    Jinxin XU, Qingwu LI, Zhiqiang GUAN, Xiaolin WANG. Nonlinear Reconstruction for Target Density Based on Randomly Perturbed Optimization and Multi-models Fusion[J]. Acta Photonica Sinica, 2022, 51(3): 0310006 Copy Citation Text show less
    First derivative products of different forward models
    Fig. 1. First derivative products of different forward models
    ACF and IACT of sample estimation for each algorithm
    Fig. 2. ACF and IACT of sample estimation for each algorithm
    The RE of sample estimation and reconstructed time by each algorithm
    Fig. 3. The RE of sample estimation and reconstructed time by each algorithm
    The RE involving multi-model fusion
    Fig. 4. The RE involving multi-model fusion
    Reconstructed results of FTO transmission image with variable density (1.25% noise level)
    Fig. 5. Reconstructed results of FTO transmission image with variable density (1.25% noise level)
    Reconstructed results of FTO transmission image with constant density (1.25% noise level)
    Fig. 6. Reconstructed results of FTO transmission image with constant density (1.25% noise level)
    High energy flash X-ray static image under 4 MeV
    Fig. 7. High energy flash X-ray static image under 4 MeV
    Results of density reconstruction (4 MeV static image)
    Fig. 8. Results of density reconstruction (4 MeV static image)
    Comparison of reconstructed density profiles of each algorithm
    Fig. 9. Comparison of reconstructed density profiles of each algorithm
    SPA

    LRIS_

    Gamma

    LRIS_

    Jeffrey

    TLE_GibbsS_RTO

    Proposed

    (T_F)

    Proposed

    (I_nF)

    Proposed

    method

    0%

    18.074 4

    /0.665 4

    19.479 2

    /0.825 4

    19.448

    /0.826 2

    20.199 5

    /0.713 9

    24.504 1

    /0.897 0

    53.267 0

    /0.999 7

    19.099 8

    /0.899 6

    25.141 0

    /0.924 7

    0.25%

    17.913 7

    /0.619 8

    19.489 6

    /0.792 8

    19.461 2

    /0.794 9

    20.284 4

    /0.695 8

    23.414 6

    / 0.886 1

    28.116 1

    /0.882 6

    19.096 9

    /0.888 4

    25.150 9

    /0.918 2

    0.75%

    16.836

    /0.528

    19.573 6

    /0.705 8

    19.554 2

    /0.704 1

    20.719 5

    /0.644 4

    23.788 1

    /0.802 2

    25.632 1

    /0.827 2

    19.030 1

    /0.821 6

    25.181 7

    /0.880 3

    1.25%

    14.583

    /0.457 9

    19.869 8

    /0.661 6

    19.850 5

    /0.660 4

    20.864 7

    /0.603 7

    22.913 6

    /0.738 3

    24.590 6

    /0.744 6

    18.870 6

    /0.750 5

    24.743 3

    /0.840 2

    Table 1. The PSNR and SSIM of reconstructed results of constant density FTO images
    SPA

    LRIS_

    Gamma

    LRIS_

    Jeffrey

    TLE_GibbsS_RTO

    Proposed

    (T_F)

    Proposed

    (I_nF)

    Proposed

    method

    0%

    19.126 7

    /0.826 5

    19.865 2

    /0.911 1

    19.833 2

    /0.910 9

    21.047 1

    /0.928 9

    26.663 7

    /0.939 6

    54.532 1

    /0.999 7

    19.661 9

    /0.950 2

    27.650 7

    /0.960 0

    0.25%

    19.037 2

    /0.787 8

    19.896 5

    /0.895

    19.845

    /0.894 6

    21.066 1

    /0.799 1

    26.714 4

    /0.927 4

    28.890 6

    /0.933 7

    19.666 7

    /0.939 5

    27.578 5

    /0.953 8

    0.75%

    18.306 9

    /0.615 2

    19.892 9

    /0.811 9

    19.835 7

    /0.807 6

    20.923 9

    /0.719 3

    25.816 5

    /0.882 4

    27.499 7

    /0.897 0

    19.587 8

    /0.872 8

    27.883 8

    /0.934 6

    1.25%

    17.001 4

    /0.469 9

    20.048

    /0.744 1

    20.012 8

    /0.741 6

    21.449 4

    /0.676 6

    25.270 6

    /0.826 1

    26.624 8

    0.864 0

    19.567 5

    /0.784 4

    27.579 1

    /0.903 6

    Table 2. The PSNR and SSIM of reconstructed results of variable density FTO images
    GPSRLRIS_JeffreyLRIS_GammaSPATLE_GibbsProposed method
    CO17.592 6/5.692 17.593 0/5.032 67.594 7/5.065 67.571 3/5.327 77.567 4/5.268 17.391 7/2.520 7
    CO27.335 3/4.737 47.398 4/2.664 47.396 4/2.661 27.367 0/2.540 57.360 5/2.431 17.308 6/2.376 5
    Table 3. Average value (g/cm3)and relative error (%)of density reconstruction of each algorithm
    Jinxin XU, Qingwu LI, Zhiqiang GUAN, Xiaolin WANG. Nonlinear Reconstruction for Target Density Based on Randomly Perturbed Optimization and Multi-models Fusion[J]. Acta Photonica Sinica, 2022, 51(3): 0310006
    Download Citation