Weld Point Cloud Segmentation Algorithm Fused Contour Feature

Abstract

In the scene of industrial quality inspection, the traditional point cloud segmentation algorithm applied to weld point cloud segmentation has the disadvantage of slow segmentation speed and low segmentation accuracy. To solve these problems, this paper proposes a weld point cloud segmentation algorithm which fuses contour features, the algorithm separates the contour and searches for key points of contour features, partitions the weld contour, then uses the RANSAC algorithm to remove the straight point cloud in each area, and finally retains the remaining point cloud and maps it back to the three-dimensional space. The experiment compared the RANSAC plane segmentation algorithm, the region growing segmentation algorithm and the improved segmentation algorithm. The experimental results show that the improved segmentation algorithm has the advantages of faster segmentation speed, higher segmentation accuracy and good robustness, and solves the problems of over-segmentation and under-segmentation in the traditional point cloud segmentation.