The Fuzzy Clustering Algorithm Based on Weighted Distance Measures for Point Cloud Segmentation

In reverse engineering, segmentation is the problem of grouping the points in the original data set into subsets each of which logically belongs to a single primitive surface. This paper presented fuzzy c-means clustering (FCM) for point cloud segmentation in reverse engineering. 8D feature vectors of points including 3D coordinates, 3D normal vectors, mean curvature and Gauss curvature were taken as input feature vectors. The weighted Euclidean distance measure was used to improve segmentation result. The segmentation method operated directly on the point cloud and could identify the inner points and border points at the same time when the segmentation was implemented, creating convenience for extracting accurately feature parameters of surface. Experiment results and comparisons with SOFM show the validity of the proposed approach.