Laser & Optoelectronics Progress, Volume. 56, Issue 13, 131501(2019)

Registration of Terrestrial Laser Scanning Data Based on Projection Distribution Entropy

Jianguo Liang1,2, Maolin Chen3, and Hong Ma1,2、*
Author Affiliations
  • 1 Chongqing Survey Institute, Chongqing 401121, China
  • 2 Chongqing Engineering Research Center of Geographic National Condition Monitoring, Chongqing 401121, China
  • 3 School of Civil Engineering, Chongqing Jiaotong University, Chongqing 400074, China
  • show less
    Figures & Tables(13)
    Point cloud distributions corresponding to different rotation parameters. (a) Correct registration;(b) incorrect registration 1; (c) incorrect registration 2; (d) point cloud distribution of correct registration; (e) point cloud distribution of incorrect registration 1; (f) point cloud distribution of incorrect registration 2
    Relationship between entropy change and rotation angle around z axis. 0° corresponds to space distribution of correct registration
    Three groups of point clouds before and after registration. (a) Dataset 1 before registration; (b) dataset 2 before registration; (c) dataset 3 before registration; (d) dataset 1 after registration; (e) dataset 2 after registration; (f) dataset 3 after registration
    Registration errors caused by view change and structural similarity and symmetry. (a) Incorrect registration caused by view change and structural symmetry; (b) incorrect registration caused by structural similarity; (c) incorrect translation caused by structural similarity; (d) rotation error caused by structural symmetry
    Registration error and time in terms of different tG. (a)(b) S2-S3 of dataset 1; (c)(d) S2-S3 of dataset 2; (e)(f) S1-S2 of dataset 3
    Density change experiments corresponding to different sampling ratios
    • Table 1. Basic information of datasets

      View table

      Table 1. Basic information of datasets

      ItemDataset 1Dataset 2Dataset 3
      ScannerReigl-VZ400Reigl-VZ400STONEX300
      Angularresolution /(°)0.040.040.08
      Point densityrange /cm0.1-60.1-70.5-13
      Averagepoint number4073886118210302081043
      Number of stations663
      Station positionacquisition methodTapemeasureMobileGPSRTK
    • Table 2. Distances between adjacent station positions and registration order

      View table

      Table 2. Distances between adjacent station positions and registration order

      SourceReferenceDataset 1Dataset 2Dataset 3
      Initialdistance /mReferencedistance /mInitialdistance /mReferencedistance /mInitialdistance /mReferencedistance /m
      S1S215.15015.12743.428.71231.51331.520
      S2S318.06018.32243.046.52937.76837.724
      S3S420.81320.79144.538.302--
      S5S417.90917.90854.447.843--
      S6S518.86417.17348.728.896--
    • Table 3. Registration error

      View table

      Table 3. Registration error

      DatasetConversion parameter errorDistance error
      Δφ /(°)Δω /(°)Δκ /(°)Δx /cmΔy /cmΔz /cmRMSD /cmStation distance error /cm
      Dataset 10.120.150.064.34.21.53.12.2
      Dataset 20.200.220.096.55.53.92.34.9
      Dataset 30.300.240.118.07.02.52.98.5
    • Table 4. Registration results of dataset 1

      View table

      Table 4. Registration results of dataset 1

      SourceReferenceOverlap rate /%MethodNumber ofsuccessful/failed runsExecution time ofcoarse registration /s
      SIFT1/18.336
      S1S252.4FPFH27/30178.3±11.2
      Proposed method20/2643.6±12.2
      SIFT0/17.678
      S2S323.1FPFH1/30169.1±4.8
      Proposed method22/2641.6±10.4
      SIFT1/18.492
      S3S435.7FPFH17/30220.5±9.6
      Proposed method18/2655.4±13.3
      SIFT0/14.651
      S5S424.9FPFH6/30211.9±7.0
      Proposed method17/2647.3±11.2
      SIFT0/18.633
      S6S525.9FPFH1/3040.1±13.4
      Proposed method11/2622.0±7.0
    • Table 5. Registration results of dataset 2

      View table

      Table 5. Registration results of dataset 2

      SourceReferenceOverlap rate /%MethodNumber ofsuccessful/failed runsExecution time ofcoarse registration /s
      SIFT0/118.7
      S1S223.7FPFH1/301470.2±28.8
      Proposed method12/26100.2±47.8
      SIFT0/117.8
      S2S326.6FPFH2/30966.6±38.7
      Proposed method20/26162.2±27.0
      SIFT0/116.9
      S3S419.3FPFH0/302587.7±153.8
      Proposed method14/26194.5±52.9
      SIFT0/113.5
      S5S429.6FPFH1/302090.2±281.5
      Proposed method15/26139.5±103.3
      SIFT0/111.2
      S6S532.1FPFH1/301543.6
      Proposed method13/26158.2±29.6
    • Table 6. Registration results of dataset 3

      View table

      Table 6. Registration results of dataset 3

      SourceReferenceOverlap rate /%MethodNumber ofsuccessful/failed runsExecution time ofcoarse registration /s
      S1S217.6
      SIFT0/148.3FPFH1/302438.3±261.7
      Proposed method19/2699.5±74.8
      SIFT0/149.127
      S2S322.9FPFH2/303477.1±318.1
      Proposed method15/2688±75.2
    • Table 7. Registration mean root errors corresponding to different tilted angles

      View table

      Table 7. Registration mean root errors corresponding to different tilted angles

      Rotation angle /(°)12345678910
      Around X axis /cm4.04.14.14.14.34.3FailureFailureFailureFailure
      Around Y axis /cm4.14.04.24.14.24.3FailureFailureFailureFailure
    Tools

    Get Citation

    Copy Citation Text

    Jianguo Liang, Maolin Chen, Hong Ma. Registration of Terrestrial Laser Scanning Data Based on Projection Distribution Entropy[J]. Laser & Optoelectronics Progress, 2019, 56(13): 131501

    Download Citation

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

    Category: Machine Vision

    Received: Jan. 14, 2019

    Accepted: Jan. 28, 2019

    Published Online: Jul. 11, 2019

    The Author Email: Ma Hong (mah@cqkcy.com)

    DOI:10.3788/LOP56.131501

    Topics