Vehicle Point Cloud Segmentation Method Based on Improved Euclidean Clustering

Abstract

Point cloud segmentation is the key technology of automatic vehicle location in factories at present. Aiming at the problem that traditional Euclidean clustering algorithm is sensitive to distance threshold and easily causes over segmentation or under segmentation of clustering objects, an improved Euclidean clustering algorithm is proposed. The improved algorithm first uses the preprocessing method to reduce the noise of the initial point cloud data, then filters the point cloud on the ground where the vehicle is parked and the environment through the random sampling consistency algorithm, and finally uses the smoothness parameter to re optimize the Euclidean clustering algorithm. The experiment applies the improved Euclidean clustering algorithm to the clustering of vehicle target point clouds. The experimental results show that the improved Euclidean clustering algorithm has a good clustering effect in a certain range of large distance threshold interval, reduces the difficulty of selecting distance threshold of traditional Euclidean clustering algorithm, for the vehicle point cloud segmentation in the case of adhesion between the head and the carriage, the accuracy is improved by about 5%, and meets the requirements of vehicle point cloud segmentation and positioning.