Optics and Precision Engineering, Volume. 32, Issue 4, 535(2024)

Urban road scenes utilize lightweight fast point cloud auto-registration of poles-like and lane lines

Huiyou ZHAO... Xuequn WU* and Yonghua XIA |Show fewer author(s)
Author Affiliations
  • College of Mechanical Engineering, Kunming University of Science and Technology, Kunming600093, China
  • show less
    Figures & Tables(14)
    Algorithm flow of this article
    Schematic diagram of tree pole and street lamp projection extraction
    Density split reflection intensity statistical chart
    Grayscale map of the lane line grid extracted by Otsu
    Schematic diagram of plane-to-plane matching
    Schematic diagram of the study area
    L001 area point cloud color map
    Lane line extraction rendering
    Pole-like extraction rendering
    Extraction result of registration primitives
    Initial estimation pairing of SAC-IA algorithm
    Registration results of different algorithms
    Comparison of convergence speeds of six algorithms
    • Table 1. Registration accuracy and run time for each algorithm

      View table
      View in Article

      Table 1. Registration accuracy and run time for each algorithm

      AlgorithmTime/sNumber of iterationsRMSE/m

      ICP

      PP-ICP

      FGR

      ISS-ICP

      Pole-ICP

      Proposed ICP

      186.231

      162.224

      859.095

      51.749

      20.138

      19.622

      200

      200

      95

      90

      70

      20

      0.537 51

      0.214 35

      0.090 39

      0.146 85

      4.861 2×10-5

      1.987 7×10-5

    Tools

    Get Citation

    Copy Citation Text

    Huiyou ZHAO, Xuequn WU, Yonghua XIA. Urban road scenes utilize lightweight fast point cloud auto-registration of poles-like and lane lines[J]. Optics and Precision Engineering, 2024, 32(4): 535

    Download Citation

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

    Category:

    Received: Aug. 11, 2023

    Accepted: --

    Published Online: Apr. 2, 2024

    The Author Email: WU Xuequn (wuxuequn520@163. com)

    DOI:10.37188/OPE.20243204.0535

    Topics