• Laser & Optoelectronics Progress
  • Vol. 56, Issue 1, 011006 (2019)
Jia Li, Ping Duan*, Yongxiang Yao, and Feng Cheng
Author Affiliations
  • College of Tourism and Geographical Sciences, Yunnan Normal University, Kunming, Yunnan 650500, China
  • show less
    DOI: 10.3788/LOP56.011006 Cite this Article Set citation alerts
    Jia Li, Ping Duan, Yongxiang Yao, Feng Cheng. Image Registration Method Based on Accelerated Segmentation Feature Optimization[J]. Laser & Optoelectronics Progress, 2019, 56(1): 011006 Copy Citation Text show less
    Template for corner detection based on Bresenham circle
    Fig. 1. Template for corner detection based on Bresenham circle
    Skewed edge pseudo corner. (a) Edge; (b) decision circle of p; (c) detection result of p
    Fig. 2. Skewed edge pseudo corner. (a) Edge; (b) decision circle of p; (c) detection result of p
    Image pyramid
    Fig. 3. Image pyramid
    Sampling points of FREAK descriptor
    Fig. 4. Sampling points of FREAK descriptor
    Image registration frame diagram of FAST combined with FREAK
    Fig. 5. Image registration frame diagram of FAST combined with FREAK
    Rotation transformation
    Fig. 6. Rotation transformation
    Scale transformation
    Fig. 7. Scale transformation
    Light transformation
    Fig. 8. Light transformation
    View transformation
    Fig. 9. View transformation
    Image mosaic results
    Fig. 10. Image mosaic results
    Reprojection errorRegistration method
    SIFT algorithmBRIEF algorithmFREAK algorithmProposed algorithm
    Number of originalimage keypoints10578531650482696
    Number of registeredimage keypoints11985467247492392
    Matching pairs2189572667144
    Inner points1507437649117
    Total projection error1592989.8051488.871116.729
    Mean projection error1.05642.26502.29410.9976
    Table 1. Reprojection error of homography pixel
    RegistrationalgorithmKeypointextractionFeaturedescriptionFeaturematchingImageregistrationWholeprocess
    SIFT algorithm0.3506141.0876532.04580.5089463.993013
    BRIEF algorithm0.0320610.8701681.05380.4697652.49071
    FREAK algorithm0.0314930.3531481.00490.4229042.07261
    Proposed algorithm0.4688600.2071200.48780.2870491.90601
    Table 2. Running time of each phase of registrationms
    Jia Li, Ping Duan, Yongxiang Yao, Feng Cheng. Image Registration Method Based on Accelerated Segmentation Feature Optimization[J]. Laser & Optoelectronics Progress, 2019, 56(1): 011006
    Download Citation