Unmanned Vehicle 3D Lidar Point Cloud Segmentation

Abstract

The solid state lidar is one of important tools for environment sensing of unmanned platform, and has been widely used in vehicle environment modeling. However, due to the low resolution, sensitive noise and complex scene, the effective segment of the whole scene is a key issue during unmanned platform data processing. In the paper, an improved 3D point clouds segmentation method is proposed for multi-line lidar in practice. After extraction building façade based on curvature segmentation, weighted Euclidean clustering is utilized to classify buildings and vegetation bodies. Then, experiments are performed on the real data acquired by the unmanned platform and the effectiveness of the proposed method is verified by comparing with the commonly used building growth segmentation algorithm.