• Journal of Applied Optics
  • Vol. 44, Issue 2, 330 (2023)
Zixiang ZHOU1, Dandan HUANG1,*, and Zhi LIU2
Author Affiliations
  • 1School of Electronical and Information Engineering, Changchun University of Science and Technology, Changchun 130000, China
  • 2Institute of Space Optoelectronics Technology, Changchun University of Science and Technology, Changchun 130000, China
  • show less
    DOI: 10.5768/JAO202344.0202005 Cite this Article
    Zixiang ZHOU, Dandan HUANG, Zhi LIU. Point cloud registration algorithm based on 3D shape context features[J]. Journal of Applied Optics, 2023, 44(2): 330 Copy Citation Text show less
    Flow chart of proposed algorithm
    Fig. 1. Flow chart of proposed algorithm
    Comparison of point cloud before and after voxel filtering
    Fig. 2. Comparison of point cloud before and after voxel filtering
    Comparison results of three commonly-used point cloud feature extraction methods
    Fig. 3. Comparison results of three commonly-used point cloud feature extraction methods
    Diagram of feature space division of 3DSC
    Fig. 4. Diagram of feature space division of 3DSC
    Matching results of bunny
    Fig. 5. Matching results of bunny
    Matching results of fr1_desk
    Fig. 6. Matching results of fr1_desk
    Matching results of fr2_desk
    Fig. 7. Matching results of fr2_desk
    Matching results of fr2_xyz
    Fig. 8. Matching results of fr2_xyz
    特征点提取方法特征点数时间/s
    ISS算法660.403
    3D-harris算法350.692
    3D-sift算法521.265
    Table 1. The number of feature points extracted by different methods
    搜索半径/m0.030.040.050.060.07
    源点云1159591444388363
    目标点云1381707582509486
    Table 2. The number of feature points extracted by ISS algorithm under different search radii
    参数数值
    ISS L2/L1的阈值0.65
    ISS L3/L2的阈值0.5
    3DSC球面最小半径/m0.02
    3DSC领域点半径/m0.1
    3DSC密度计算阈值0.02
    Table 3. Other main parameters of proposed algorithm
    数据集点云原始数量下采样后的点云数量特征点个数粗匹配前的匹配点对粗匹配后的匹配点对匹配成功率/%
    bunny348342996219141285.7
    348252918721
    fr1_desk189053158565444726286.1
    186134156961582
    fr2_desk159187135065119161487.5
    146533124076147
    fr2_xyz10115811601741060413585.3
    10752882373011381
    Table 4. Number of point clouds and matching of feature points
    配准方法时间/s均方根误差/m
    本文算法0.0414.70×10−7
    SAC-IA+ICP算法1.0237.09×10−4
    ISS+3DSC+ndt算法0.0945.46×10−5
    Table 5. Evaluation of registration effect of bunny
    配准方法时间/s均方根误差/m
    本文算法0.2292.00×10−4
    SAC-IA+ICP算法11.4034.54×10−4
    ISS+3DSC+ndt算法0.2313.05×10−4
    Table 6. Evaluation of registration effect of fr1_desk
    配准方法时间/s均方根误差/m
    本文算法0.0632.31×10−4
    SAC-IA+ICP算法8.2234.50×10−4
    ISS+3DSC+ndt算法0.0743.02×10−4
    Table 7. Evaluation of registration effect of fr2_desk
    配准方法时间/s均方根误差/m
    本文算法2.8071.68×10−5
    SAC-IA+ICP算法14.9764.91×10−4
    ISS+3DSC+ndt算法2.8872.99×10−5
    Table 8. Evaluation of registration effect of fr2_xyz
    Zixiang ZHOU, Dandan HUANG, Zhi LIU. Point cloud registration algorithm based on 3D shape context features[J]. Journal of Applied Optics, 2023, 44(2): 330
    Download Citation