DOI: 10.1166/jno.2024.3572 ISSN: 1555-130X

An Improved RANSAC Based Segmentation Algorithm for the Belt Wedge Point Cloud of the Automotive V-Ribbed Belt

Wu-Sheng Tang, Yun-Fei Guo, Chun-Mei Yin, Jin-Hua Zhang, Yao-Chen Shi, Qing-Hua Li
  • Electrical and Electronic Engineering
  • Electronic, Optical and Magnetic Materials

Point cloud segmentation is an important part of the belt wedge dimension measurement of automotive V-ribbed belt, and accurate segmentation is the basis of the subsequent dimension calculation of the belt wedge. In view of the problem of randomly selecting initial points and over-segmented plane in orthodox RANSAC algorithm, this text presents an modified RANSAC arithmetic for segmentation of belt wedge point clouds of automotive V-ribbed belt, which uses the radius space to redefine the selection method of initial points, obtains the optimal plane result through multiple iterations, and uses the angle and distance of vectors to prevent over-segmentation of point clouds. In the experiment, RANSAC and improved RANSAC are used to segment the belt wedge point cloud datas of cloud data of the PK type automotive V-ribbed belt collected by line laser displacement sensor. Compare segmentation results, Facts prove that the improved RANSAC algorithm has better segmentation effect, the proportion of plane points h is increased by 7.5%.

More from our Archive