• Acta Photonica Sinica
  • Vol. 50, Issue 7, 285 (2021)
Linmei ZHU1, Xiucheng DONG1, Zhengyu ZHANG2, Fan ZHANG1, Haibin WANG1, and Lei REN1
Author Affiliations
  • 1School of Electrical and Electronic Information, Xihua University, Chengdu60039, China
  • 2High Speed Aerodynamics Institute, China Aerodynamics Research and Development Center, Mianyang, Sichuan61000, China
  • show less
    DOI: 10.3788/gzxb20215007.0715001 Cite this Article
    Linmei ZHU, Xiucheng DONG, Zhengyu ZHANG, Fan ZHANG, Haibin WANG, Lei REN. Camera Pose Estimation Algorithm for Singular Configuration of Target Points[J]. Acta Photonica Sinica, 2021, 50(7): 285 Copy Citation Text show less
    Camera imaging model
    Fig. 1. Camera imaging model
    Schematic diagram of PST algorithm
    Fig. 2. Schematic diagram of PST algorithm
    Center of gravity position diagram
    Fig. 3. Center of gravity position diagram
    Central position diagram
    Fig. 4. Central position diagram
    Simulation result with respect to varying numbers from 4 to 15 in ordinary-3D point configuration
    Fig. 5. Simulation result with respect to varying numbers from 4 to 15 in ordinary-3D point configuration
    Simulation result with respect to varying numbers from 4 to 15 in quasi-linear point configuration
    Fig. 6. Simulation result with respect to varying numbers from 4 to 15 in quasi-linear point configuration
    Simulation result with respect to varying numbers from 4 to 15 in planar point configuration
    Fig. 7. Simulation result with respect to varying numbers from 4 to 15 in planar point configuration
    Simulation result with respect to varying noise levels from 0.5 to 5 in case of ordinary-3D configuration when point number n=4
    Fig. 8. Simulation result with respect to varying noise levels from 0.5 to 5 in case of ordinary-3D configuration when point number n=4
    Simulation result with respect to varying noise levels from 0.5 to 5 in case of ordinary-3D configuration when point number n=5
    Fig. 9. Simulation result with respect to varying noise levels from 0.5 to 5 in case of ordinary-3D configuration when point number n=5
    Experimental environment configuration
    Fig. 10. Experimental environment configuration
    The target points of pose experiment
    Fig. 11. The target points of pose experiment
    Re-projection error of image plane
    Fig. 12. Re-projection error of image plane
    NumbersOur method/sIEPnP/sEPnP+GN/sLHM/s
    n=200.165 40.165 20.324 60.446 5
    n=600.223 50.272 00.365 41.278 8
    n=1000.292 70.308 40.404 12.101 0
    n=2000.446 40.633 40.498 74.065 4
    Table 1. Running time of different algorithms
    AlgorithmError (4 points)/mmError (5 points)/mmError(20 points)/mm
    PlanarQuasi⁃linearPlanarQuasi⁃linearPlanar
    IEPnP2.2383.0521.7922.5093.040
    EPnP0.3240.1460.0050.0360.004
    LHM0.0620.0360.0580.0390.044
    Our method0.0030.0020.0020.0020.002
    Table 2. The average result of re-projection error
    Linmei ZHU, Xiucheng DONG, Zhengyu ZHANG, Fan ZHANG, Haibin WANG, Lei REN. Camera Pose Estimation Algorithm for Singular Configuration of Target Points[J]. Acta Photonica Sinica, 2021, 50(7): 285
    Download Citation