A Contour-Feature Based on Method of Point Segmentation for Sagger
Hao Deng, Jianqi An, Kang Yu, Jianru Xiong · 2025
In the industrial process of robotic sagger gripping, the accuracy of point cloud segmentation of sagger directly determines the accuracy of sagger positioning, which affects the final gripping result. However, the existing point cloud segmentation methods for sagger have insufficient segmentation accuracy and do not fully use the contour features of sagger. This paper presents a RANSAC point cloud segmentation method based on contour features of sagger. First, the method realizes the segmentation of the sagger contour by separating the sagger contour and searching the key points of the contour features. Then, the segmentation result of the sagger contour is divided into multiple regions, and the RANSAC algorithm is used to remove the curve point cloud in each region. Finally, the remaining sagger point cloud is mapped back to the 3D space, which improves the sagger localization accuracy. The experiments compare the performance of existing methods with the method in this paper from the perspective of time complexity and accuracy, and make full use of the contour features of sagger to prove that the method has the advantages of high segmentation accuracy, fast segmentation speed, and high robustness, which solves the problems of over-segmentation and under-segmentation in conventional point cloud segmentation.