基于最优匹配的三维CAD模型相似性评价算法

Optimal-matching-based 3D CAD model similarity assessment algorithm

提出一种基于最优匹配的三维计算机辅助设计模型相似性评价算法。根据三维计算机辅助设计模型的边界表示,将相比较的两个计算机辅助设计模型拆分成面,并分别对面进行分类。将两模型中同种类型的面两两匹配,运用相应的准则计算出每对面的相似性系数iδj,从而使同种类型的面形成了一个二分图。利用图论中二分图的最优匹配算法,计算出两个模型同类型面中各面的最优匹配方案,并计算出加权最优匹配值,最后将各种类型面的最优匹配值相加得到模型的相似系数,以此作为评价模型相似性的依据。为恰当地评价两个类型相同面的相似性,给出了能够综合反映面几何特征和拓扑特征的评价准则。试验表明本方法能够客观地反映两计算机辅助设计模型的相似程度,并较好地反映人视觉上的相似经验。

An algorithm for 3D Computer Aided Design （CAD） model similarity assessment was presented, Boundary Representation （B-Rep） Models were decomposed into the surfaces. For each model, according to the surface＇s type, the surfaces were stored into the different sets. Then, the surface-pairs sets were formed in which each surface-pair contained two surfaces coming from the different models with the same type. Coefficient of similarity of the two surfaces for each pair was computed by using some rules. For same type of surface, there were two sets belonging to the two comparative models and thus a bi-graph was obtained, Then, Kuhn-Munkres algorithm was used to seek a matching of maximum total weight from the weighted of the bi-graph, The total weight which is processed is the value of the optimal solution. For each type of surface, the value of the optimal solution was obtained. The sum of all types＇ values was regarded as the similar degree of the two models, To correctly assess the similarity of the two surfaces which were belonging to same type, some rules influenced the geometric features and the topological features of the models were given . Experiment showed that this algorithm could effectively give the similar degree of the two models and reflect human perceptual similarity.