Journal of Applied Optics, Volume. 46, Issue 4, 805(2025)

Point cloud registration algorithm based on feature extraction and improved ICP

Ruiyu WANG1,2、*, Huicheng JING1,2, and Qilong BAO1,2
Author Affiliations
  • 1School of Electrical Engineering, North China University of Science and Technology, Tangshan 063000, China
  • 2Hebei Wind-Solar-Hydrogen-Storage System Security Monitoring and Intelligent Operation Center of Technology Innovation, Tangshan 063000, China
  • show less
    Figures & Tables(13)
    Flow chart of algorithm
    Schematic diagram of OCtree
    Schematic diagram of local coordinate system
    Schematic diagram of FPFH[14]
    Comparison of voxel down-sampling time
    Point-cloud registration results
    Detailed diagrams of registration effect of proposed algorithm
    Comparison of registration results of different algorithms
    • Table 1. Comparison of point cloud quantity before and after voxel down-sampling

      View table
      View in Article

      Table 1. Comparison of point cloud quantity before and after voxel down-sampling

      点云数目BunnyDragonHappy BuddhaOffice
      滤波前点云数目35947437645543652223276
      体素下采样后点云数目1080311057115399105
      Octree加速的体素下采样后点云数目1082311069116479154
    • Table 2. Bunny rotation errors

      View table
      View in Article

      Table 2. Bunny rotation errors

      配准算法旋转误差/rad
      XYZ
      经典ICP0.1337250.0233410.098621
      3DSC+ICP1.73479×10−31.70325×10−30.016454
      文献[16]2.38122×10−20.1110850.345353
      文献[17]3.10758×10−21.71309×10−31.74167×10−2
      本文算法2.80863×10−41.46857×10−48.9103×10−4
    • Table 3. Dragon rotation errors

      View table
      View in Article

      Table 3. Dragon rotation errors

      配准算法旋转误差/rad
      XYZ
      经典ICP3.8583×10−57.93745×10−50.00370514
      3DSC+ICP3.34844×10−30.0168310.025531
      文献[16]1.84899×10−23.82977×10−20.110419
      文献[17]5.89098×10−49.26305×10−41.19257×10−3
      本文算法6.32509×10−55.92008×10−57.62284×10−5
    • Table 4. Happy Buddha rotation errors

      View table
      View in Article

      Table 4. Happy Buddha rotation errors

      配准算法旋转误差/rad
      XYZ
      经典ICP0.0171500.0197260.0030386
      3DSC+ICP0.0334490.0298950.0023111
      文献[16]1.756650.8608020.901115
      文献[17]4.72674×10−25.28701×10−27.40129×10−3
      本文算法2.11837×10−56.66402×10−53.54886×10−4
    • Table 5. Office rotation errors

      View table
      View in Article

      Table 5. Office rotation errors

      配准算法旋转误差/rad
      XYZ
      经典ICP0.1747550.2640860.626325
      3DSC+ICP0.0273210.0610327.76237×10−3
      文献[16]0.1508730.3402360.167988
      文献[17]0.5032970.4018660.698001
      本文算法7.17528×10−51.57768×10−53.07625×10−4
    Tools

    Get Citation

    Copy Citation Text

    Ruiyu WANG, Huicheng JING, Qilong BAO. Point cloud registration algorithm based on feature extraction and improved ICP[J]. Journal of Applied Optics, 2025, 46(4): 805

    Download Citation

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

    Category:

    Received: Feb. 29, 2024

    Accepted: --

    Published Online: Sep. 16, 2025

    The Author Email: Ruiyu WANG (wry18251780190@163.com)

    DOI:10.5768/JAO202546.0402003

    Topics