Laser & Optoelectronics Progress, Volume. 58, Issue 6, 610002(2021)

Independent Method for Selecting Radius of FPFH Neighborhood in 3D Point Cloud Registration

Zhao Mingfu1,2, Cao Libo1,3, Song Tao1,2, Liu Shuai1, Luo Yuhang1, and Yang Xin1
Author Affiliations
  • 1College of Electrical and Electronic Engineering, Chongqing University of Technology, Chongqing 400054, China
  • 2Chongqing University Engineering Center of Elevator Intelligent Operation and Maintenance, Chongqing 402260, China
  • 3Chongqing Key Laboratory of Optical Fiber Sensing and Photoelectric Detection, Chongqing 400054, China
  • show less
    Figures & Tables(12)
    Optimization algorithm for FPFH feature extraction in point cloud registration
    Schematic of circular density calculation
    Local coordinate system
    Schematic of FPFH influence range
    Fitting function between neighborhood radius and registration error under different circular densities
    The optimal function
    Registration results in bunny dataset. (a) Original point cloud and target point cloud; (b) fixed FPFH neighborhood radius; (c) selected FPFH neighborhood radius by the optimal function
    Registration results in dragonStandRight dataset. (a) Original point cloud and target point cloud; (b) fixed FPFH neighborhood radius; (c) selected FPFH neighborhood radius by the optimal function
    Registration results in happyStandRight dataset. (a) Original point cloud and target point cloud; (b) fixed FPFH neighborhood radius; (c) selected FPFH neighborhood radius by the optimal function
    • Table 1. Registration results of different algorithms in bunny dataset

      View table

      Table 1. Registration results of different algorithms in bunny dataset

      Algorithmk /mICP score /10-6TFPFH+TICP /s
      Manual regulation0.0029.7739014.413+148.463
      0.0069.7164318.774+74.650
      0.0109.8297114.414+79.920
      0.0309.7863014.352+31.745
      0.0509.7727014.238+28.162
      0.0809.8494915.508+41.871
      Proposed algorithmAutomatic selection9.7264815.956+27.636
    • Table 2. Registration results of different algorithms in dragonStandRight dataset

      View table

      Table 2. Registration results of different algorithms in dragonStandRight dataset

      Algorithmk /mICP score /10-6TFPFH+TICP /s
      Manual regulation0.0023.2106716.325+157.907
      0.0063.2249015.727+73.216
      0.0103.2059916.439+75.906
      0.0303.2244019.590+45.493
      0.0503.2033219.463+29.827
      0.0803.1966020.920+32.208
      Proposed algorithmAutomatic selection3.1941517.470+31.801
    • Table 3. Registration results of different algorithms in happyStandRight dataset

      View table

      Table 3. Registration results of different algorithms in happyStandRight dataset

      Algorithmk /mICP score /10-6TFPFH+TICP /s
      Manual regulation0.0026.5335612.620+117.283
      0.0066.573569.713+286.572
      0.0106.5380011.390+72.669
      0.0306.5641512.687+108.061
      0.0506.5351812.068+78.406
      Proposed algorithmAutomatic selection6.5373412.349+62.589
    Tools

    Get Citation

    Copy Citation Text

    Zhao Mingfu, Cao Libo, Song Tao, Liu Shuai, Luo Yuhang, Yang Xin. Independent Method for Selecting Radius of FPFH Neighborhood in 3D Point Cloud Registration[J]. Laser & Optoelectronics Progress, 2021, 58(6): 610002

    Download Citation

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

    Category: Image Processing

    Received: Jul. 24, 2020

    Accepted: --

    Published Online: Mar. 11, 2021

    The Author Email:

    DOI:10.3788/LOP202158.0610002

    Topics