面向三维CAD模型检索的模型分割方法

摘要

In order to promote the effectiveness of model segmentation,an improved method of model segmentation for 3D CAD model retrieval is proposed in this paper.Firstly,3D CAD models are transformed to attribute adjacency labelled graphs (AALGs) based on the boundary representation (B-Rep) information such as geometry and topology attributes.Secondly,the models are decomposed into surface regions initially according to the convexity-concavity of nodes and links in AALGs.The segmented sub-graphs corresponding to the decomposed regions are merged to constructthe optimal region decomposition which makes the defined segmentation cohesion be maximal.When there are several optional combined sub-graphs,the optimal one is selected and merged according to the maximal region coupling factors.Finally,the similarity between the models is assessed by optimal-matching of decomposed regions.The experimental results show that the present method of model segmentation is effective for 3D model retrieval so as to support the effective reuse of CAD model.%为了提高三维CAD模型分割的效果及其对模型检索支持的有效性,提出一种面向三维CAD模型检索的模型分割方法.首先通过提取B-rep形式CAD模型的几何拓扑属性,将模型转换为属性邻接标记图;然后依据图中节点和连线的凸凹性将模型初步分割为凸凹性连续的局部区域,再以分割内聚度最大为目标,基于区域耦合度进行可合并区域的选择与优化合并;最后采用区域最优匹配的方法进行模型间的相似性比较和检索.实验结果表明,该模型分割方法能有效支持模型检索,可实现CAD模型的重用.