Volume XLII-4/W18
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-4/W18, 1117–1122, 2019
https://doi.org/10.5194/isprs-archives-XLII-4-W18-1117-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-4/W18, 1117–1122, 2019
https://doi.org/10.5194/isprs-archives-XLII-4-W18-1117-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

  19 Oct 2019

19 Oct 2019

CLASSIFICATION OF MOBILE TERRESTRIAL LIDAR POINT CLOUD IN URBAN AREA USING LOCAL DESCRIPTORS

M. Zaboli1, H. Rastiveis1, A. Shams2, B. Hosseiny1, and W. A. Sarasua3 M. Zaboli et al.
  • 1Department of Photogrammetry and Remote Sensing, School of Surveying and Geospatial Engineering, College of Engineering, University of Tehran, Tehran, Iran
  • 2Advanced Highway Maintenance & Construction Technology (AHMCT) Research Center, University of California, Davis, CA, USA
  • 3Glenn Department of Civil Engineering, Clemson University, Clemson, SC, USA

Keywords: Mobile Terrestrial LiDAR, Point Cloud, Classification, Geometric Features, Photogrammetry

Abstract. Automated analysis of three-dimensional (3D) point clouds has become a boon in Photogrammetry, Remote Sensing, Computer Vision, and Robotics. The aim of this paper is to compare classifying algorithms tested on an urban area point cloud acquired by a Mobile Terrestrial Laser Scanning (MTLS) system. The algorithms were tested based on local geometrical and radiometric descriptors. In this study, local descriptors such as linearity, planarity, intensity, etc. are initially extracted for each point by observing their neighbor points. These features are then imported to a classification algorithm to automatically label each point. Here, five powerful classification algorithms including k-Nearest Neighbors (k-NN), Gaussian Naive Bayes (GNB), Support Vector Machine (SVM), Multilayer Perceptron (MLP) Neural Network, and Random Forest (RF) are tested. Eight semantic classes are considered for each method in an equal condition. The best overall accuracy of 90% was achieved with the RF algorithm. The results proved the reliability of the applied descriptors and RF classifier for MTLS point cloud classification.