An extraction method for interested buildings using lidar point clouds data

Abstract

LiDAR (Light Detection and Ranging) is an active remote sensing technique for acquiring spatial information. It can quickly acquire three-dimensional (3D) geographic coordinate information of ground surface and ground targets, and has typical advantage in such applications as urban planning, 3D modeling, disaster assessment, etc. This paper presents an extraction method for interested buildings using three-dimensional laser point cloud data which are filtered and organized by the kd tree. First, the algorithm determines candidate points of a building from non-ground points and clusters them on the constraints of distance so that single building target can be segmented. Second, for each segmented building target, the algorithm extracts its edge points and regularizes its edge. The extracted building feature information is provided for quickly searching target of interest. At last, the method is proved to be effective based on the analysis of measured data. The method is no point cloud interpolation error, and is not affected by the size or shape of a building.