节点文献
基于层次聚类的网格模型自动分割方法
Automatic mesh segmentation based on hierarchical clustering
【摘要】 将网格模型按照表面特征进行分割,在计算机图形学中有着广泛的应用。在传统的网格分割层次聚类算法的基础上,利用相邻聚类公共边的长度,公共顶点的曲率,以及聚类的面积对聚类之间的合并成本进行加权,提出新的合并成本函数,更好地利用网格表面特征指导聚类之间的合并过程。同时,针对传统分割算法中分割区域边界存在的锯齿现象,针对每一个分割区域,全局考虑其与相邻分割区域的模糊地带,利用最大流最小割的方法优化分割区域边界,从而消除锯齿,平滑边界。
【Abstract】 Automatic mesh segmentation has been widely used in computer graphics.Based on traditional hierarchical clustering segmentation methods,a new cost function was presented which combined the length of common boundary of adjacent clusters,the curvature of common vertices and the area of each cluster.The new cost function could more clearly represent the properties of the mesh,and perform better in the clusters merging process.Besides,considering the jagged common boundaries,a flow network was constructed in the fuzzy area of each cluster and all of its adjacent ones to optimize and smooth the boundaries by minimum cut algorithm.
【Key words】 surface feature; automatic segmentation; hierarchical clustering; merging cost function; minimum cut;
- 【文献出处】 计算机应用 ,Journal of Computer Applications , 编辑部邮箱 ,2010年S2期
- 【分类号】TP391.41
- 【被引频次】5
- 【下载频次】162