Research And Implementation Of Model Segmentation Based On Feature Points

The goal of mesh segmentation is to mark each grid vertex as a particular region.How to quickly and accurately separate mesh is a focus on the segmentation algorithm.This paper presents an algorithm of mesh segmentation based on feature points, whichcan quickly and effectively divide three-dimensional grid model.Feature point extraction algorithm based on the shortest distance, first obtains thegeometric center of the model vr. According to vr, calculating the shortest distanceon the model of the two largest peak vs1and vs2. Let E1be the set of local extremes tothe source vertex vs1, and E2be the set of local extremes to the source vertex vs2. Theintersection of E1and E2is the set of feature points.With wide spread problem of thegeneral mesh segmentation algorithm, such as curvature calculation complex and fuzzydivided borders. This paper presents a curvature calculation method by the averagevector of the first neighborhood of vertex, and use curvature of the margin adjacent totwo points as the regional growth factor of Fast Marching watershed algorithm.Through experiment, the proposed method can segment the models perfectly bothin time and accuracy. For the feature point extraction part, with the help of userinteraction, we can get a much more stable result in the number of the segmentationparts. In addition, curvature computation of the new method is simple. And through onetime separating operation, it can gain relative clear boundaries. At last, we haveachieved a simple but useful model segmentation method.