• Acta Photonica Sinica
  • Vol. 52, Issue 4, 0410004 (2023)
Zefeng HUANG, Shen YANG*, Huiping DENG, and Qingson LI
Author Affiliations
  • School of Information Science and Engineering, Wuhan University of Science and Technology, Wuhan 430081, China
  • show less
    DOI: 10.3788/gzxb20235204.0410004 Cite this Article
    Zefeng HUANG, Shen YANG, Huiping DENG, Qingson LI. Light Field All-in-focus Image Fusion Based on MDLatLRR and KPCA[J]. Acta Photonica Sinica, 2023, 52(4): 0410004 Copy Citation Text show less
    Digital refocusing parameterization model of light field image
    Fig. 1. Digital refocusing parameterization model of light field image
    Full-focus image fusion algorithm flow of light field based on multi-scale latent low-rank decomposition
    Fig. 2. Full-focus image fusion algorithm flow of light field based on multi-scale latent low-rank decomposition
    Results of three-scale latent low-rank decomposition
    Fig. 3. Results of three-scale latent low-rank decomposition
    Local step difference weighting method
    Fig. 4. Local step difference weighting method
    Multi-scale iterative significance detection algorithm
    Fig. 5. Multi-scale iterative significance detection algorithm
    Score of salient matrix under different iterations
    Fig. 6. Score of salient matrix under different iterations
    Framework of eigencoefficient matrix fusion algorithm based on kernel principal component analysis
    Fig. 7. Framework of eigencoefficient matrix fusion algorithm based on kernel principal component analysis
    Four groups of light field refocusing images and corresponding focusing decision map
    Fig. 8. Four groups of light field refocusing images and corresponding focusing decision map
    The total focusing decision map of each refocusing image and the corresponding all-in-focus image are displayed
    Fig. 9. The total focusing decision map of each refocusing image and the corresponding all-in-focus image are displayed
    LoveLocks image fusion results comparison
    Fig. 10. LoveLocks image fusion results comparison
    Edelweiss image fusion results comparison
    Fig. 11. Edelweiss image fusion results comparison
    SF↑Var↑QcvQcbQabfMI↑EN↑EI↑CE↓AG↑
    RW9.0946.3192.780.730.7755.277.08932.020.09829.28
    f-PCNN9.4948.90212.270.680.7564.387.18034.110.01131.22
    IFCNN9.0948.47137.860.590.6024.057.17532.750.01429.95
    MGFF8.7751.02412.630.590.7053.717.14833.110.09230.34
    MST9.5449.04111.510.700.7624.637.16434.050.01431.15
    CSR9.0848.3490.760.690.7334.877.14532.150.01029.46
    M&P9.5150.2892.730.770.7635.657.15133.190.00830.38
    Table 1. Objective index comparison of LoveLocks image on seven fusion algorithms
    SF↑Var↑QcvQcbQabfMI↑EN↑EI↑CE↓AG↑
    RW21.8071.1384.850.800.8384.897.95577.660.08869.26
    f-PCNN22.0270.65118.670.750.8213.377.96379.110.07070.54
    IFCNN22.0168.36210.740.560.6932.967.94877.960.05769.57
    MGFF20.0372.81154.580.640.7572.757.93071.890.10564.20
    MST21.9570.3596.310.720.8233.137.91878.130.10069.71
    CSR21.7871.9987.150.810.8174.427.96378.120.07469.55
    M&P22.0972.6988.850.870.8395.517.96479.100.07670.46
    Table 2. Objective index comparison of Edelweiss images on seven fusion algorithms
    SF↑Var↑QcvQcbQabfMI↑EN↑EI↑CE↓AG↑
    RW10.4760.3438.600.7760.7656.147.55137.230.036 833.98
    f-PCNN10.6160.4981.210.7240.7355.057.55237.910.026 434.61
    IFCNN10.3860.2772.740.5940.5914.657.54536.520.021 833.36
    MGFF9.8662.36169.860.6340.6814.347.54335.430.062 332.39
    MST10.6860.7248.590.7400.7515.207.54237.890.031 934.59
    CSR10.3560.6534.420.7220.7145.537.55136.280.021 533.12
    M&P10.7560.7935.920.7950.7616.277.55637.930.021 434.59
    Table 3. Average objective index comparison of seven fusion algorithms
    AlgorithmRWf-PCNNIFCNNMGFFMSTCSRM&P
    Time cost22.51532.1411.853.530.05228.81204.18
    Table 4. Average time cost of seven fusion algorithms
    Zefeng HUANG, Shen YANG, Huiping DENG, Qingson LI. Light Field All-in-focus Image Fusion Based on MDLatLRR and KPCA[J]. Acta Photonica Sinica, 2023, 52(4): 0410004
    Download Citation