Adaptive segmentation method based on similarity of laser point cloud topology

Abstract

Laser point cloud segmentation is the basis of target splicing or recognition. In this paper, a point cloud segmentation method based on point topology is proposed. The relationship between neighboring points is obtained by the curvature relationship between points. The relationship within point cloud between each point is established, and then the point cloud is divided by cutting the graph. The feature of eigenvalue of the Laplacian matrix realizes the adaptive segmentation. Three different kind of point cloud are tested with the algorithm in this paper and the result show that the algorithm has good performance on point cloud cutting of obvious characteristics and robust to noise.