Chinese Journal of Lasers, Volume. 45, Issue 11, 1104004(2018)

An Automatic Segmentation Algorithm for Dense Pipeline Point Cloud Data

Huang Kai1, Cheng Xiaojun1,2, Jia Dongfeng3, Hu Danhua4, and Hu Minjie5
Author Affiliations
  • 1[in Chinese]
  • 2[in Chinese]
  • 3[in Chinese]
  • 4[in Chinese]
  • 5[in Chinese]
  • show less
    Figures & Tables(21)
    Flow chart of algorithm
    Interior view of a power plant
    Schematic of octree structure
    Comparison between (a) global fitting and (b) local fitting
    Schematic of plane point cloud filtering based on normal vector constraints
    Schematic of plane filtering and spatial clustering
    Schematic of pipeline data merging
    Virtual scenes in experiment 1. (a) Top view; (b) front view; (c) left view; (d) southeast side view
    Point cloud data in experiment 1
    Sketch after removal of large planes in experiment 1
    Segmentation results of pipeline data in experiment 1. (a) Southeast side view; (b) top view
    Relationship among station locations in experiment 2
    Grayscale image of station Gr01270 in experiment 2
    Sketch of registered point cloud in experiment 2. (a) Left view; (b) southeast side view
    Sketch after removal of large planes in experiment 2. (a) Top view; (b) southeast side view
    Segmentation result of pipeline data in experiment 2. (a) Top view; (b) front view; (c) southeast side view; (d) southwest side view
    Segmentation sketch of pipeline data in experiment 1, where segmented pipeline data are indicated in white domain. (a) Top view; (b) southwest side view
    Segmentation sketch of pipeline data in experiment 2, where segmented pipeline data are indicated in white domain. (a) Top view; (b) southeast side view
    • Table 1. Parameters designed for experiments

      View table

      Table 1. Parameters designed for experiments

      Parameteretθtdt1 /mw1 /%dt2 /mEminvβRmindt3 /mw2 /%w3 /%
      Experiment 1200645°0.0130.14000.54000.012525
      Experiment 2200845°0.02530.15000.511°3800.0154730
    • Table 2. Runtime of sequential and parallel segmentations of point cloud

      View table

      Table 2. Runtime of sequential and parallel segmentations of point cloud

      No.Sequentialruntime /sParallel runtime /s
      2 workers4 workers8 workers
      Experiment 112864
      Experiment 2151297
    • Table 3. Analysis of segmentation results in experiment 2

      View table

      Table 3. Analysis of segmentation results in experiment 2

      ParameterSegmentation resultParameterSegmentation result
      Ratio of true positive /%92.5Precision ratio /%94.1
      Ratio of false positive /%5.9Recall ratio /%93.9
      Ratio of false negative /%7.5
    Tools

    Get Citation

    Copy Citation Text

    Huang Kai, Cheng Xiaojun, Jia Dongfeng, Hu Danhua, Hu Minjie. An Automatic Segmentation Algorithm for Dense Pipeline Point Cloud Data[J]. Chinese Journal of Lasers, 2018, 45(11): 1104004

    Download Citation

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

    Category: Measurement and metrology

    Received: Apr. 18, 2018

    Accepted: --

    Published Online: Nov. 15, 2018

    The Author Email:

    DOI:10.3788/CJL201845.1104004

    Topics