• Acta Photonica Sinica
  • Vol. 50, Issue 9, 0910003 (2021)
Hao CHEN1, Huicheng LAI1、2, Guxue GAO1, Hao WU1, and Xuze QIAN1
Author Affiliations
  • 1College of Information Science and Engineering, Xinjiang University, Urumqi830046, China
  • 2Key Laboratory of Signal Detection and Processing, Xinjiang University, Urumqi830046, China
  • show less
    DOI: 10.3788/gzxb20215009.0910003 Cite this Article
    Hao CHEN, Huicheng LAI, Guxue GAO, Hao WU, Xuze QIAN. Sand-dust Image Enhancement Based on Multi-exposure Image Fusion[J]. Acta Photonica Sinica, 2021, 50(9): 0910003 Copy Citation Text show less
    Algorithm flow chart
    Fig. 1. Algorithm flow chart
    The histogram of sand-dust image before and after blue channel compensation and the histogram of corresponding image standardization
    Fig. 2. The histogram of sand-dust image before and after blue channel compensation and the histogram of corresponding image standardization
    The enhancement results of the proposed algorithm to the sand-dust image (Fig. 2(a)) at different values of α
    Fig. 3. The enhancement results of the proposed algorithm to the sand-dust image (Fig. 2(a)) at different values of α
    The enhancement results of the proposed algorithm to the sand image at different values of AB(A=0.5)
    Fig. 4. The enhancement results of the proposed algorithm to the sand image at different values of ABA=0.5)
    The enhancement results of the proposed algorithm to the sand image at different values of A(AB=0.5)
    Fig. 5. The enhancement results of the proposed algorithm to the sand image at different values of AAB=0.5)
    Exposure images generated by the proposed algorithm and fusion result
    Fig. 6. Exposure images generated by the proposed algorithm and fusion result
    Comparison of algorithm results (Ⅰ)
    Fig. 7. Comparison of algorithm results (Ⅰ)
    Comparison of algorithm results (Ⅱ)
    Fig. 8. Comparison of algorithm results (Ⅱ)
    Comparison of algorithm results (Ⅲ)
    Fig. 9. Comparison of algorithm results (Ⅲ)
    Comparison of algorithm results (Ⅳ)
    Fig. 10. Comparison of algorithm results (Ⅳ)
    Comparison of algorithm results (Ⅴ)
    Fig. 11. Comparison of algorithm results (Ⅴ)
    Comparison of algorithm results (Ⅵ)
    Fig. 12. Comparison of algorithm results (Ⅵ)
    Orig.Ref.[4Ref.[13Ref.[1Ref.[3Ref.[10Ref.[8Ours
    Fig.74.126 84.004 96.110 67.793 26.005 65.935 08.353 49.176 2
    Fig.81.504 32.879 12.687 44.006 22.488 02.685 24.791 95.605 6
    Fig.91.652 72.413 93.300 54.580 04.472 63.309 94.780 56.366 8
    Fig.102.560 92.401 45.571 75.051 74.512 94.187 24.613 96.063 3
    Fig.114.392 04.195 86.931 08.029 27.255 85.664 57.776 69.558 9
    Fig.122.074 72.548 93.045 14.531 63.882 24.939 24.594 86.331 2
    Average2.718 63.074 04.607 75.665 34.769 54.453 55.818 57.183 7
    Table 1. Comparison of average gradient
    Orig.Ref.[4Ref.[13Ref.[1Ref.[3Ref.[10Ref.[8Ours
    Fig.715.325 614.512 520.658 327.810 123.087 022.870 530.658 931.381 8
    Fig.84.121 67.669 46.565 010.166 06.656 57.265 412.748 514.211 0
    Fig.95.577 37.222 011.130 914.865 715.739 710.886 715.419 220.810 1
    Fig.109.569 88.255 520.079 119.013 717.827 416.394 116.946 322.551 8
    Fig.1113.744 713.753 921.023 924.644 923.756 119.155 725.715 628.868 5
    Fig.126.647 07.792 68.604 513.608 012.583 015.798 913.866 318.065 0
    Average9.164 39.867 614.676 918.351 416.608 315.395 219.225 822.648 0
    Table 2. Comparison of spatial frequency
    Orig.Ref.[4Ref.[13Ref.[1Ref.[3Ref.[10Ref.[8Ours
    Fig.7117.616 8105.468 3213.710 6387.295 3266.914 8261.930 7470.707 5493.164 5
    Fig.88.499 929.431 121.565 651.711 822.170 726.412 581.321 3101.049 5
    Fig.915.574 926.114 962.034 7110.647 7124.041 659.342 7119.041 6216.830 7
    Fig.1045.886 634.147 8202.006 4181.138 3159.239 0134.663 4143.888 9254.821 7
    Fig.1194.700 194.826 5221.568 3304.463 8282.898 9183.940 3331.492 5417.762 1
    Fig.1222.133 430.420 537.090 092.765 979.317 5125.041 396.321 2163.483 9
    Average50.735 353.401 5126.329 3188.003 8155.763 8131.888 5207.128 8274.518 7
    Table 3. Comparison of contrast
    ProcessComplexityProcessComplexity
    Blue channel compensationO(mn)Construction of Gaussian pyramidO(Nmn)
    Image standardizationO(mn)Construction of Laplacian pyramidO(Nmn)
    Generation of exposure imagesO(Nmn)Image pyramid fusion and reconstructionO(Nmn)
    Calculation of weight mapsO(Nmn)Proposed methodO(Nmn)
    Table 4. Time complexity of the proposed algorithm
    ResolutionRef.[4Ref.[13Ref.[1Ref.[3Ref.[10Ref.[8Ours
    Fig.7893×5135.403.022.511.454.934.292.40
    Fig.81 600×1 20022.1210.848.636.7018.9923.5312.41
    Fig.9900×6006.293.632.741.625.034.312.97
    Fig.10600×4002.911.911.390.982.682.261.33
    Fig.11490×3262.041.431.090.851.881.991.22
    Fig.12640×4423.451.991.661.062.942.571.95
    Total-42.2122.8218.0212.6636.4538.9522.28
    Table 5. Comparison of running time (s)
    Hao CHEN, Huicheng LAI, Guxue GAO, Hao WU, Xuze QIAN. Sand-dust Image Enhancement Based on Multi-exposure Image Fusion[J]. Acta Photonica Sinica, 2021, 50(9): 0910003
    Download Citation