Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Acta Optica Sinica, Volume. 39, Issue 3, 0315007(2019)

Scale Point Cloud Registration Algorithm in High-Dimensional Orthogonal Subspace Mapping

Yue Jiang1, Hongguang Huang1、*, Qin Shu1, Zhao Song2, and Zhirong Tang3
Author Affiliations
  • 1 School of Electrical Engineering and Information, Sichuan University, Chengdu, Sichuan 610065, China
  • 2 Southwest Institute of Technical Physics, Chengdu, Sichuan 610041, China
  • 3 College of Nuclear Technology and Automation Engineering, Chengdu University of Technology,Chengdu, Sichuan 610059, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (20)
    Equations (0)
    References (0)
    Cited By (5)
    Tools
    Paper Information
    Topics
    Figures & Tables(20)
    Flow chart of algorithm

    Fig. 1. Flow chart of algorithm

    Download full size
    Initial state of point cloud with noise and without data loss. (a) Bunny; (b) Dragon

    Fig. 2. Initial state of point cloud with noise and without data loss. (a) Bunny; (b) Dragon

    Download full size
    Registration results of Bunny obtained by different algorithms with noise and without data loss. (a) OrthS; (b) GA+ICP; (c) GA+Scale-ICP; (d) CPD; (e) Go-ICP

    Fig. 3. Registration results of Bunny obtained by different algorithms with noise and without data loss. (a) OrthS; (b) GA+ICP; (c) GA+Scale-ICP; (d) CPD; (e) Go-ICP

    Download full size
    Registration results of Dragon obtained by different algorithms with noise and without data loss. (a) OrthS; (b) GA+ICP; (c) GA+Scale-ICP; (d) CPD; (e) Go-ICP

    Fig. 4. Registration results of Dragon obtained by different algorithms with noise and without data loss. (a) OrthS; (b) GA+ICP; (c) GA+Scale-ICP; (d) CPD; (e) Go-ICP

    Download full size
    Registration results of large-scale Dragon by different algorithms with Gaussian white noise of 50 dB. (a) OrthS; (b) GA+ICP; (c) GA+Scale-ICP; (d) CPD; (e) Go-ICP

    Fig. 5. Registration results of large-scale Dragon by different algorithms with Gaussian white noise of 50 dB. (a) OrthS; (b) GA+ICP; (c) GA+Scale-ICP; (d) CPD; (e) Go-ICP

    Download full size
    Initial state of point cloud with noise and data loss. (a) Bunny; (b) Dragon

    Fig. 6. Initial state of point cloud with noise and data loss. (a) Bunny; (b) Dragon

    Download full size
    Registration results of Bunny obtained by different algorithms with noise and data loss. (a) OrthS; (b) OrthS+ICP; (c) OrthS+Scale-ICP; (d) CPD; (e) Go-ICP

    Fig. 7. Registration results of Bunny obtained by different algorithms with noise and data loss. (a) OrthS; (b) OrthS+ICP; (c) OrthS+Scale-ICP; (d) CPD; (e) Go-ICP

    Download full size
    Registration results of Dragon obtained by different algorithms with noise and data loss. (a) OrthS; (b) OrthS+ICP; (c) OrthS+Scale-ICP; (d) CPD; (e) Go-ICP

    Fig. 8. Registration results of Dragon obtained by different algorithms with noise and data loss. (a) OrthS; (b) OrthS+ICP; (c) OrthS+Scale-ICP; (d) CPD; (e) Go-ICP

    Download full size
    Registration effects under different noise environments. (a) 25 dB; (b) 20 dB; (c) 15 dB; (d) 10 dB

    Fig. 9. Registration effects under different noise environments. (a) 25 dB; (b) 20 dB; (c) 15 dB; (d) 10 dB

    Download full size
    Registration results. (a) Root mean square error; (b) registration time

    Fig. 10. Registration results. (a) Root mean square error; (b) registration time

    Download full size
    Initial state of affine point cloud. (a) Bunny; (b) Dragon

    Fig. 11. Initial state of affine point cloud. (a) Bunny; (b) Dragon

    Download full size
    Registration results of affine point cloud. (a)(c) OrthS; (b)(d) Scale-ICP

    Fig. 12. Registration results of affine point cloud. (a)(c) OrthS; (b)(d) Scale-ICP

    Download full size
    Two sets of physical maps. (a) Cylinder; (b) shower gel

    Fig. 13. Two sets of physical maps. (a) Cylinder; (b) shower gel

    Download full size
    Registration results of cylinder. (a) Initial state; (b) OrthS; (c) OrthS+ICP; (d) OrthS+Scale-ICP; (e) CPD; (f) Go-ICP

    Fig. 14. Registration results of cylinder. (a) Initial state; (b) OrthS; (c) OrthS+ICP; (d) OrthS+Scale-ICP; (e) CPD; (f) Go-ICP

    Download full size
    Registration results of shower gel. (a) Initial state; (b) OrthS; (c) OrthS+ICP; (d) OrthS+Scale-ICP; (e) CPD; (f) Go-ICP

    Fig. 15. Registration results of shower gel. (a) Initial state; (b) OrthS; (c) OrthS+ICP; (d) OrthS+Scale-ICP; (e) CPD; (f) Go-ICP

    Download full size

    • Table 1. Comparison of point cloud registration data with noise and without data loss

      View table

      Table 1. Comparison of point cloud registration data with noise and without data loss

      AlgorithmTime /sRMSE/mm
      BunnyDragonBunnyDragon
      OrthS3.65.30.47230.0008
      GA+ICP407.0330.50.47170.0008
      GA+Scale-ICP14.830.70.79670.0410
      CPD174.889.70.43540.0033
      Go-ICP50.727.70.90050.0220

    • Table 2. Comparison of point cloud registration data with Gaussian white noise of 50 dB

      View table

      Table 2. Comparison of point cloud registration data with Gaussian white noise of 50 dB

      AlgorithmOrthSGA+ICPGA+Scale-ICPCPDGo-ICP
      Time /s33.56243.9177.6454.137.0
      RMSE/mm6.821×10-40.00560.01250.00290.0388

    • Table 3. Comparison of point cloud registration data with noise and data loss

      View table

      Table 3. Comparison of point cloud registration data with noise and data loss

      AlgorithmTime /sR'MSE /mm
      BunnyDragonBunnyDragon
      OrthS2.84.200.47230.0011
      OrthS+ICP218.284.400.47140.0011
      OrthS+Scale-ICP5.99.890.67230.0048
      CPD133.770.100.40460.0035
      Go-ICP82.025.901.46910.0196

    • Table 4. Comparison of registration data between two registration algorithms

      View table

      Table 4. Comparison of registration data between two registration algorithms

      AlgorithmTime /sR'MSE /mm
      BunnyDragonBunnyDragon
      OrthS3.03.90.22050.0208
      Scale-ICP13.98.30.97300.0309

    • Table 5. Comparison of point cloud registration data

      View table

      Table 5. Comparison of point cloud registration data

      AlgorithmTime /sR'MSE /mm
      CylinderShower gelCylinderShower gel
      OrthS4.72.60.63682.9621
      OrthS+ICP184.2378.80.62442.9378
      OrthS+Scale-ICP7.14.90.642413.9816
      CPD36.157.03.08361.9082
      Go-ICP26.627.00.58173.0128
    Tools

    Get Citation

    Copy Citation Text

    Yue Jiang, Hongguang Huang, Qin Shu, Zhao Song, Zhirong Tang. Scale Point Cloud Registration Algorithm in High-Dimensional Orthogonal Subspace Mapping[J]. Acta Optica Sinica, 2019, 39(3): 0315007

    Download Citation

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

    Category: Machine Vision

    Received: Oct. 22, 2018

    Accepted: Nov. 19, 2018

    Published Online: May. 10, 2019

    The Author Email:

    DOI:10.3788/AOS201939.0315007

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology