Building Boundary Extraction from LiDAR Point Cloud Data
Emon Kumar Dey, Mohammad Awrangjeb, Fayez Tarsha Kurdi, Bela Stantić
Abstract
Building boundary extraction from LiDAR point cloud data is important for urban planning and 3D modelling. Due to the uneven point distribution, missing data, and occlusion in LiDAR point cloud data, extraction of boundary points is challenging. Existing approaches have shortcomings either in detecting boundary points on concave shapes or separate identification of ‘hole’ boundary points inside the building roof. This paper, presents a method for detecting both inner and outer boundary points of the extracted building point cloud. Based on the properties of Delaunay Triangulation and distance from the mean point of the calculated neighbourhood for any point, we extract both inner and outer boundary points. Experimental results using some synthetic shapes as well as some real datasets show the competitive performance of the proposed method.