• Laser & Optoelectronics Progress
  • Vol. 58, Issue 22, 2228009 (2021)
Yangping Wang1、2、3, Xibing Liu1、*, Jingyu Yang1, and Jianwu Dang1、3
Author Affiliations
  • 1School of Electronics and Information Engineering, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China
  • 2Experimental Teaching Center on Computer Science, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China
  • 3Gansu Provincial Engineering Research Center for Artificial Intelligence and Graphic & Image Processing, Lanzhou, Gansu 730070, China;
  • show less
    DOI: 10.3788/LOP202158.2228009 Cite this Article Set citation alerts
    Yangping Wang, Xibing Liu, Jingyu Yang, Jianwu Dang. Dense Matching of Multi-View Remote Sensing Terrain Image Based on Improved PMVS Algorithm[J]. Laser & Optoelectronics Progress, 2021, 58(22): 2228009 Copy Citation Text show less
    Reconstruction process of PMVS algorithm
    Fig. 1. Reconstruction process of PMVS algorithm
    Reconstruction process of improved PMVS algorithm
    Fig. 2. Reconstruction process of improved PMVS algorithm
    Process flow chart of concurrent SIFT operator based on image block
    Fig. 3. Process flow chart of concurrent SIFT operator based on image block
    Schematic diagram of image segmentation strategy
    Fig. 4. Schematic diagram of image segmentation strategy
    Constraint relationship between feature point f and potential candidate matching point f'
    Fig. 5. Constraint relationship between feature point f and potential candidate matching point f'
    Detected results of features. (a) Four terrain images of features to be detected; (b) detected results of features by Harris operator; (c) detected results of features by SIFT operator; (d) detected results of features by improving SIFT operator
    Fig. 6. Detected results of features. (a) Four terrain images of features to be detected; (b) detected results of features by Harris operator; (c) detected results of features by SIFT operator; (d) detected results of features by improving SIFT operator
    Reconstruction results of two areas using PMVS algorithm and proposed algorithm. (a) Images to be reconstructed; (b) reconstruction results of PMVS algorithm; (c) reconstruction results of proposed algorithm
    Fig. 7. Reconstruction results of two areas using PMVS algorithm and proposed algorithm. (a) Images to be reconstructed; (b) reconstruction results of PMVS algorithm; (c) reconstruction results of proposed algorithm
    Reconstruction results of two areas using PMVS algorithm and proposed algorithm. (a) Images to be reconstructed; (b) reconstruction results of PMVS; (c) reconstruction results of ours
    Fig. 8. Reconstruction results of two areas using PMVS algorithm and proposed algorithm. (a) Images to be reconstructed; (b) reconstruction results of PMVS; (c) reconstruction results of ours
    Image resolutionFeature point extraction time /s
    Harris operatorSIFT operatorImproved SIFT operator
    2457×20000.9861742.7421341.028954
    2451×20000.9091851.8576540.897063
    1765×17680.5414611.2586140.695861
    2451×20000.9953642.8145871.085487
    Table 1. Time performance comparison of Harris operator, SIFT operator, and improved SIFT operator
    Image resolutionNORP of PMVSNORP of improved PMVSPER /%Time of PMVS /sTime of improved PMVS /sTime increase rate /%
    2457×2000159894166922~4.410.9589.587~12.5
    2451×2000210251222081~5.610.7269.191~14.3
    1765×1768189487201693~6.49.6248.587~10.8
    2451×2000217687232025~6.610.9869.486~13.7
    Table 2. Comparison of reconstruction performance between original PMVS algorithm and improved PMVS algorithm
    Yangping Wang, Xibing Liu, Jingyu Yang, Jianwu Dang. Dense Matching of Multi-View Remote Sensing Terrain Image Based on Improved PMVS Algorithm[J]. Laser & Optoelectronics Progress, 2021, 58(22): 2228009
    Download Citation