• Acta Photonica Sinica
  • Vol. 52, Issue 2, 0211006 (2023)
Jiang DING1, Chaocheng SONG1,2, Cui MA2, Kai HE2, and Qiyang ZUO2,*
Author Affiliations
  • 1College of Mechanical Engineering,Guangxi University,Nanning 530004,China
  • 2Shenzhen Institute of Advanced Technology,Chinese Academy of Sciences,Shenzhen 518055,China
  • show less
    DOI: 10.3788/gzxb20235202.0211006 Cite this Article
    Jiang DING, Chaocheng SONG, Cui MA, Kai HE, Qiyang ZUO. Monocular Vision Pose Solving Algorithm Based on Feature Circle Target[J]. Acta Photonica Sinica, 2023, 52(2): 0211006 Copy Citation Text show less
    Characteristic circular target
    Fig. 1. Characteristic circular target
    Camera imaging principle
    Fig. 2. Camera imaging principle
    perspective projection principle
    Fig. 3. perspective projection principle
    Schematic projection of a circular target
    Fig. 4. Schematic projection of a circular target
    Variation of a1 and a2 lengths under perspective distortion
    Fig. 5. Variation of a1 and a2 lengths under perspective distortion
    The variation of absolute error accuracy of ellipse method
    Fig. 6. The variation of absolute error accuracy of ellipse method
    Absolute error variation of different algorithms(r=15 mm)
    Fig. 7. Absolute error variation of different algorithms(r=15 mm)
    Experimental platform of spatial plane pose measurement
    Fig. 8. Experimental platform of spatial plane pose measurement
    Imaging region of camera at different object distances
    Fig. 9. Imaging region of camera at different object distances
    Experiments of absolute error based on different algorithms(r=15 mm)
    Fig. 10. Experiments of absolute error based on different algorithms(r=15 mm)
    Equipment nameModelPrecision parameterQuantity
    Industrial robotABB IRB1200-7/0.7Repositioning accuracy 0.02 mm1
    Industry camera

    HIKVISION MV-CE200-10GM 20

    million pixels static black-and-white camera

    Resolution 5 472 × 3 648 ;

    Pixel dimension 2.4 µm × 2.4 µm

    1
    LensHIKVISION MVL-KF2528M-12MP(neutral medium)

    Fixed focus 25 mm;manual operture;

    12 million pixels;FA lens;field of view

    (D-H-V): 37.6°-30.4°-23°

    1
    Adjustable ring light sourceJL-FLA-AR1206048 V/624 W1
    Lighting controllerJL-APR241
    Calibration board7×750 mm×50 mm;±0.001 mm1
    Table 1. Main equipment
    Object distance/mmd/rAlgorithmMaximum error/(°)Minimum error/(°)Mean value(0°~85°)/(°)

    Quadratic mean

    deviation(0°~85°)/(°)

    Precision/(°)
    138.89.25Algorithm12.182 30.008 40.3500.4541
    Our algorithm0.491 80.002 10.0870.1080.1
    Algorithm22.433 00.001 30.2520.4621
    175.011.67Algorithm11.294 40.006 70.2310.3121
    Our algorithm0.752 20.000 20.0860.1440.1
    Algorithm22.738 10.002 00.2160.4261
    266.217.75Algorithm11.252 60.002 80.1500.2871
    Our algorithm1.118 80.000 20.0820.1960.1
    Algorithm23.724 60.001 00.4060.7961
    Table 2. Results of three algorithms(r=15 mm)
    Jiang DING, Chaocheng SONG, Cui MA, Kai HE, Qiyang ZUO. Monocular Vision Pose Solving Algorithm Based on Feature Circle Target[J]. Acta Photonica Sinica, 2023, 52(2): 0211006
    Download Citation