Optics and Precision Engineering, Volume. 33, Issue 13, 2136(2025)

Key feature registration of point cloud normal vector and curvature

Zhenchen JI1,2,3, Hongxu AI1,2,3, Yuan HAN1,3, Jiaqi YAO1,3, Youzhi LI1, Yanqiu WANG1,3, Fu ZHENG1,3, Wenjie WANG2, and Zhibin SUN1,3、*
Author Affiliations
  • 1National Space Science Center, Chinese Academy of Sciences, Beijing0090, China
  • 2North China Electric Power University, Beijing1006, China
  • 3University of Chinese Academy of Sciences, Beijing100049, China
  • show less
    Figures & Tables(25)
    Flowchart of point cloud registration algorithm for geometric feature
    Key point feature descriptor
    Flowchart of RANSAC algorithm
    Flowchart of ICP algorithm
    Initial point cloud
    Key point extraction effect
    Relationship between number of key points and radius and threshold
    Select effect of key points under different threshold
    Curvature comparison between key points and non-key points
    Point cloud processing effect
    Key point extraction effect
    Correspondence optimization process
    Correspondence optimization curve
    Curve of root mean square error variation
    Registration and fusion effect
    Principle diagram of time-of-flight three-dimensional area array detection system
    Three-dimensional at point cloud reconstruction object
    Registration fusion results
    Noise resistance test
    Angle and registration error at different Gaussian noise levels
    • Table 1. Key points extract parameters and results

      View table
      View in Article

      Table 1. Key points extract parameters and results

      模型点云数量关键点数量优化率(%)
      Bunny40 2565 82514.47
      Dragon41 8416 45815.43
      Happy78 05615 03519.26
    • Table 2. Statistics of key point quantity

      View table
      View in Article

      Table 2. Statistics of key point quantity

      NumberPoints cloudKey pointsRate/%
      17 0354 04523.75
      16 7243 72422.27
      10°16 5523 69522.32
      15°21 3186 09128.57
      20°16 1103 73723.20
      25°15 9483 48121.83
      30°15 7793 60122.82
      35°15 6163 69023.63
      40°16 1743 88324.01
      45°15 8113 36121.26
      50°15 0873 63124.07
    • Table 3. Registration parameters

      View table
      View in Article

      Table 3. Registration parameters

      CountC1C2Transformation matrix
      17 0357015901.0000.090
      01.0000
      16 724-0.0901.000
      111.31.462.111
    • Table 4. Registration data comparison

      View table
      View in Article

      Table 4. Registration data comparison

      ModelGO-ICPNDTFeature+RANSACOurs
      RMSE/mmTime/sRMSE/mmTime/sRMSE/mmTime/sRMSE/mmTime/s
      Mr. LiPei(0°-5°)6.6623.366.971.5017.105.333.326.01
      Mr. LiPei(0°-50°)33.5019.6131.571.7029.936.4215.597.23
    • Table 5. Registration data comparison

      View table
      View in Article

      Table 5. Registration data comparison

      配准方法旋转误差/(°)RMSE/mmTime/s
      XYZ
      F+R-2.739-6.190-29.6654.843.67
      GO-ICP0.002-1.0070.0493.282.27
      NDT0.002-1.00203.665.60
      Ours0.1760.427-0.0712.714.19
    Tools

    Get Citation

    Copy Citation Text

    Zhenchen JI, Hongxu AI, Yuan HAN, Jiaqi YAO, Youzhi LI, Yanqiu WANG, Fu ZHENG, Wenjie WANG, Zhibin SUN. Key feature registration of point cloud normal vector and curvature[J]. Optics and Precision Engineering, 2025, 33(13): 2136

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category:

    Received: Apr. 1, 2025

    Accepted: --

    Published Online: Aug. 28, 2025

    The Author Email:

    DOI:10.37188/OPE.20253313.2136

    Topics