Bulletin of Surveying and Mapping ›› 2020, Vol. 0 ›› Issue (9): 80-84.doi: 10.13474/j.cnki.11-2246.2020.0288

Previous Articles     Next Articles

Progressive building boundary extraction from original LiDAR point cloud data

WEN Ping1, ZHAO Fei1,2, WU Xiaodong1, WANG Ying1, WANG Chong1   

  1. 1. Kunming Survey, Design and Research Institute Co., Ltd., of China Power Construction Group, Kunming 650051, China;
    2. School of Geosciences, Yunnan University, Kunming 650091, China
  • Received:2019-11-26 Revised:2020-01-12 Online:2020-09-25 Published:2020-09-28

Abstract: Building is an important element of urban 3D modeling, and the extraction of building boundary information is both difficult and important. In this paper, a method is proposed to extract the regular building boundary from the original LiDAR point cloud data: Firstly, it uses the progressive mathematical morphology filter to separate the non ground points from the original point cloud data. Then it uses the improved 3D Hough transformation to classify the non ground points. On this basis, the key points are determined to simplify and fit the building boundary according to the azimuth threshold of the adjacent points. Finally, the building boundary is regularized based on the length weighted direction of the boundary. The results show that this method greatly improves the efficiency and accuracy of point cloud processing, and can automatically and incrementally obtain building boundary information directly from the initial data. At the same time, this method has a good effect on solving the problems of small building volume, close distance and large roof slope in small and medium-sized towns.

Key words: building boundary, original LiDAR point cloud, 3D Hough transformation, boundary simplification, length weighted regularization

CLC Number: