• Laser & Optoelectronics Progress
  • Vol. 59, Issue 12, 1211002 (2022)
Ziwen Yu, Ning Zhang*, Yue Pan**, Yue Zhang, and Yuxuan Wang
Author Affiliations
  • School of Opto-Electronic Engineering, Changchun University of Science and Technology, Changchun 130022, Jilin , China
  • show less
    DOI: 10.3788/LOP202259.1211002 Cite this Article Set citation alerts
    Ziwen Yu, Ning Zhang, Yue Pan, Yue Zhang, Yuxuan Wang. Heterogeneous Image Matching Based on Improved SIFT Algorithm[J]. Laser & Optoelectronics Progress, 2022, 59(12): 1211002 Copy Citation Text show less
    Flow chart of proposed algorithm
    Fig. 1. Flow chart of proposed algorithm
    Improved extremum detection
    Fig. 2. Improved extremum detection
    Process of coordinate rotation
    Fig. 3. Process of coordinate rotation
    Feature points of visible light image
    Fig. 4. Feature points of visible light image
    Feature points of infrared image
    Fig. 5. Feature points of infrared image
    Feature points of visible image. (a) SIFT algorithm;(b)proposed algorithm
    Fig. 6. Feature points of visible image. (a) SIFT algorithm;(b)proposed algorithm
    Feature points of infrared image. (a) SIFT algorithm; (b) proposed algorithm
    Fig. 7. Feature points of infrared image. (a) SIFT algorithm; (b) proposed algorithm
    Distribution of 9 subregions
    Fig. 8. Distribution of 9 subregions
    Group 1 experimental data. (a) Visible light image; (b) infrared image
    Fig. 9. Group 1 experimental data. (a) Visible light image; (b) infrared image
    Processing results of first group of experimental data by different algorithms. (a)‒(c) Connection diagrams when SIFT algorithm, ASIFT algorithm, and proposed algorithm are matched; (d) transform of infrared and visible images and contour overlap of matching results of proposed algorithm
    Fig. 10. Processing results of first group of experimental data by different algorithms. (a)‒(c) Connection diagrams when SIFT algorithm, ASIFT algorithm, and proposed algorithm are matched; (d) transform of infrared and visible images and contour overlap of matching results of proposed algorithm
    Group 2 experimental data. (a) Visible light image; (b) infrared image
    Fig. 11. Group 2 experimental data. (a) Visible light image; (b) infrared image
    Processing results of second group of experimental data by different algorithms. (a)‒(c) Connection diagrams when SIFT algorithm, ASIFT algorithm, and proposed algorithm are matched; (d) transform of infrared and visible images and contour overlap of matching results of proposed algorithm
    Fig. 12. Processing results of second group of experimental data by different algorithms. (a)‒(c) Connection diagrams when SIFT algorithm, ASIFT algorithm, and proposed algorithm are matched; (d) transform of infrared and visible images and contour overlap of matching results of proposed algorithm
    Group 3 experimental data. (a) Visible light image; (b) infrared image
    Fig. 13. Group 3 experimental data. (a) Visible light image; (b) infrared image
    Processing results of third group of experimental data by different algorithms. (a)‒(c) Connection diagrams when SIFT algorithm, ASIFT algorithm, and proposed algorithm are matched; (d) transform of infrared and visible images and contour overlap of matching results of proposed algorithm
    Fig. 14. Processing results of third group of experimental data by different algorithms. (a)‒(c) Connection diagrams when SIFT algorithm, ASIFT algorithm, and proposed algorithm are matched; (d) transform of infrared and visible images and contour overlap of matching results of proposed algorithm
    Group 4 experimental data. (a) Visible light image; (b) infrared image
    Fig. 15. Group 4 experimental data. (a) Visible light image; (b) infrared image
    Processing results of fourth group of experimental data by different algorithms. (a)‒(c) Connection diagrams when SIFT algorithm, ASIFT algorithm, and proposed algorithm are matched; (d) transform of infrared and visible images and contour overlap of matching results of proposed algorithm
    Fig. 16. Processing results of fourth group of experimental data by different algorithms. (a)‒(c) Connection diagrams when SIFT algorithm, ASIFT algorithm, and proposed algorithm are matched; (d) transform of infrared and visible images and contour overlap of matching results of proposed algorithm
    Experimental dataAlgorithmNumber of feature pointsCorrect matching pointRMSE /pixelTime /s
    Reference imagesImages to be registered
    Data 1SIFT55631781844.327.83
    ASIFT2350022727435.819.941
    Proposed algorithm548716421351.526.58
    Data 2SIFT2319820102.317.893
    ASIFT106048982484.211.833
    Proposed algorithm2258918631.315.62
    Data 3SIFT5994181310Mismatch28.32
    ASIFT2112014279294.815.602
    Proposed algorithm532113691851.224.88
    Data 4SIFT529912532Mismatch23.38
    ASIFT22112153420Mismatch15.761
    Proposed algorithm510211241161.723.42
    Table 1. Comparison of registration results
    Ziwen Yu, Ning Zhang, Yue Pan, Yue Zhang, Yuxuan Wang. Heterogeneous Image Matching Based on Improved SIFT Algorithm[J]. Laser & Optoelectronics Progress, 2022, 59(12): 1211002
    Download Citation