Building Planar Surfaces Segmentation in LiDAR Data Using Adaptive Mean Shift Algorithm

This paper presents an algorithm for the segmentation of building planar surfaces from airborne laser data. Planar surface structure analysis is fundamental to almost any application involving LiDAR data, especially building model reconstruction. The proposed algorithm estimates feature vector of each laser foot print in a neighbourhood, and cluster the planar surface using a robust unsupervised cluster method—adaptive mean shift (AMS). This algorithm is general as it aims at extracting palanr surface segments that exhibit an homogeneous behavior with restriction to one specific pattern, it doesn’t need any priori knowledge about planar numbers and their parameters. The algorithm adopts a more complex feature space definition for this purpose, which offers a very general and flexible way to identify homogeneous patterns in the data. Efficiency of the algorithm is proved by experiments.