• Laser & Optoelectronics Progress
  • Vol. 62, Issue 2, 0237001 (2025)
Xinggui Xu1,*, Hong Li1, Bing Ran2, Weihe Ren3, and Junrong Song1
Author Affiliations
  • 1The School of Information, Yunnan University of Finance and Economics, Kunming 650051, Yunnan , China
  • 2Institute of Optics and Electronics, Chinese Academy of Sciences, Chengdu 610054, Sichuan , China
  • 3The Institute of Beijing Space Electromechanical Research, Beijing 100039, China
  • show less
    DOI: 10.3788/LOP240707 Cite this Article Set citation alerts
    Xinggui Xu, Hong Li, Bing Ran, Weihe Ren, Junrong Song. Turbulence-Blurred Target Restoration Algorithm with a Nonconvex Regularization Constraint[J]. Laser & Optoelectronics Progress, 2025, 62(2): 0237001 Copy Citation Text show less
    The framework of proposed algorithm
    Fig. 1. The framework of proposed algorithm
    Component results of LatLRSD. (a)‒(d) Original turbulence blur image XO, low rank dictionary structural componentXZ*, hidden component L*X, and noise component E*
    Fig. 2. Component results of LatLRSD. (a)‒(d) Original turbulence blur image XO, low rank dictionary structural componentXZ*, hidden component L*X, and noise component E*
    Iterative results of eigenvalues for LatLRSD. (a)‒(c) Singular value results corresponding to the noise component, low rank component, and hidden component in the low rank space decomposition at iteration numbers of #5, #25, #50, and #100
    Fig. 3. Iterative results of eigenvalues for LatLRSD. (a)‒(c) Singular value results corresponding to the noise component, low rank component, and hidden component in the low rank space decomposition at iteration numbers of #5, #25, #50, and #100
    Schematic diagrams of regularization constraint conditions. (a) MC constraint; (b) scaled GMC constraint
    Fig. 4. Schematic diagrams of regularization constraint conditions. (a) MC constraint; (b) scaled GMC constraint
    The turbulence-degraded targets restoration results of non-convex regularization. (a)‒(b) Restoration results for non-convex objective functions in two different scenarios at iteration numbers of #5, #25, #50, and #100
    Fig. 5. The turbulence-degraded targets restoration results of non-convex regularization. (a)‒(b) Restoration results for non-convex objective functions in two different scenarios at iteration numbers of #5, #25, #50, and #100
    Original images. (a)‒(d) Flow field image, infrared turbulence image, long-range turbulence distortion image, and synthesized turbulence blur image
    Fig. 6. Original images. (a)‒(d) Flow field image, infrared turbulence image, long-range turbulence distortion image, and synthesized turbulence blur image
    Restoration results of algorithm of reference [12]
    Fig. 7. Restoration results of algorithm of reference [12]
    Restoration results of algorithm of reference [16]
    Fig. 8. Restoration results of algorithm of reference [16]
    Restoration results of algorithm of reference [18]
    Fig. 9. Restoration results of algorithm of reference [18]
    Restoration results of algorithm of reference [20]
    Fig. 10. Restoration results of algorithm of reference [20]
    Restoration results of proposed algorithm
    Fig. 11. Restoration results of proposed algorithm
    AlgorithmMESDICEISNR /dB
    Algorithm of reference [127.50930.08120.24920.70829.2
    Algorithm of reference [168.10240.04530.29150.40986.4
    Algorithm of reference [188.24630.05050.30640.43576.2
    Algorithm of reference [207.26550.04010.20270.51608.5
    Proposed algorithm8.71360.08870.39830.780410.3
    Table 1. Restoration evaluation results of Fig.6(a)
    AlgorithmMESDICEISNR /dB
    Algorithm of reference [127.40640.01520.22040.33324.7
    Algorithm of reference [167.77120.01610.18640.38325.5
    Algorithm of reference [188.05840.01870.22350.43217.7
    Algorithm of reference [207.02310.01590.13170.20013.5
    Proposed algorithm8.15060.01970.23450.53348.0
    Table 2. Restoration evaluation results of Fig.6(b)
    AlgorithmMESDICEISNR /dB
    Algorithm of reference [125.13340.01860.20150.39572.8
    Algorithm of reference [165.81540.01120.19640.33456.5
    Algorithm of reference [186.18700.01070.23540.36408.5
    Algorithm of reference [204.50870.01790.23690.34653.6
    Proposed algorithm6.49050.02360.33620.465411.1
    Table 3. Restoration evaluation results of Fig.6(c)
    AlgorithmMESDICEISNR /dB
    Algorithm of reference [127.53360.02060.22190.19413.7
    Algorithm of reference [165.46320.02140.27630.24577.5
    Algorithm of reference [185.61270.02100.28540.19808.1
    Algorithm of reference [205.08700.02200.20010.19676.2
    Proposed algorithm6.09040.03080.35720.23507.2
    Table 4. Restoration evaluation results of Fig.6(d)
    Xinggui Xu, Hong Li, Bing Ran, Weihe Ren, Junrong Song. Turbulence-Blurred Target Restoration Algorithm with a Nonconvex Regularization Constraint[J]. Laser & Optoelectronics Progress, 2025, 62(2): 0237001
    Download Citation