Segmentation Algorithm for 3D LiDAR Point Cloud Based on Region Clustering

Abstract

Environmental perception is a key technology in autonomous cars and mobile robotics system. The clustering and segmentation of the point cloud data obtained by sensors is an important step to realize environmental perception. In order to solve the shortcomings of over-segmentation in some current segmentation algorithms, a fast segmentation algorithm for 3D LIDAR cloud points is proposed. The Ground Plane Fitting algorithm is used to segment the ground surface, remove the ground cloud points interference, and combine the spatial Euclidean distance parameter with the angle parameter of the adjacent scan line of the LIDAR to cluster and segment the non-ground cloud points. The experimental result show that compared with some traditional algorithms, this method can effectively reduce the segmentation error rate, improve the accuracy of non-ground target segmentation, and significantly improve the segmentation efficiency.