Laser & Optoelectronics Progress, Volume. 59, Issue 4, 0428004(2022)

Airborne LiDAR Point Cloud Filtering Method Based on Multiconstrained Connected Graph Segmentation

Zhenyang Hui*, Haiying Hu, Na Li, and Zhuoxuan Li
Author Affiliations
  • Faculty of Geomatics, East China University of Technology, Nanchang , Jiangxi 330013, China
  • show less
    Figures & Tables(16)
    Flow chart of point cloud filtering
    Diagram of verticality
    Diagram of height difference constraint and distance condition constraint
    Diagram of connected graph. (a) Connected graphs without constraints; (b) connected graphs with constraints
    Two dimensional grid diagram of seed points
    Add new ground seed points to the blank grid
    Diagram of distance from point to fitted plane
    Point cloud data filtering results. (a) DSM of raw data; (b) true DEM; (c) DEM of the filtering results of the proposed method; (d) error distribution of filtering results of the proposed method
    Comparison of total errors of the five methods
    Comparison of mean values of the three kinds of error of the five methods
    Average total errors of different verticality thresholds
    Average total errors of different height difference thresholds
    Average total errors of different distance thresholds
    • Table 1. 15 groups of point cloud data and their characteristics

      View table

      Table 1. 15 groups of point cloud data and their characteristics

      EnvironmentSiteSampleFeature
      City111Hillsides, low vegetation, buildings
      12Hillsides, buildings
      221Large buildings, bridges
      22Irregular structure
      23Large irregular structure
      24Steep sides
      331Complex building complex
      441Blank data
      42Train tracks and trains
      Country551Steep sides, low vegetation, blank data
      52
      53
      54
      661Roads, buildings, data gaps
      771Bridges, roads, an underground passage
    • Table 2. Error matrix

      View table

      Table 2. Error matrix

      CategoryFiltering result
      Number of ground pointsNumber of object points
      Reference resultNumber of ground pointsab
      Number of object pointscd
    • Table 3. Filtering error of 15 experiment data of the proposed method

      View table

      Table 3. Filtering error of 15 experiment data of the proposed method

      SampleT1 /%T2 /%Ttotal /%ckappa
      Average5.729.295.440.81
      Sample 1126.7413.2621.040.58
      Sample 127.921.394.760.90
      Sample 212.712.942.760.92
      Sample 223.2115.607.030.83
      Sample 238.004.336.270.87
      Sample 246.1010.067.150.82
      Sample 310.371.851.050.98
      Sample 411.014.632.750.94
      Sample 4211.100.433.550.91
      Sample 511.336.732.420.92
      Sample 523.0917.354.410.75
      Sample 536.3619.346.840.44
      Sample 542.463.913.240.94
      Sample 613.477.513.570.56
      Sample 711.9330.034.810.72
    Tools

    Get Citation

    Copy Citation Text

    Zhenyang Hui, Haiying Hu, Na Li, Zhuoxuan Li. Airborne LiDAR Point Cloud Filtering Method Based on Multiconstrained Connected Graph Segmentation[J]. Laser & Optoelectronics Progress, 2022, 59(4): 0428004

    Download Citation

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

    Category: Remote Sensing and Sensors

    Received: Mar. 30, 2021

    Accepted: Apr. 14, 2021

    Published Online: Jan. 25, 2022

    The Author Email: Zhenyang Hui (huizhenyang2008@163.com)

    DOI:10.3788/LOP202259.0428004

    Topics