中文摘要：

提出了一种基于字符串度量的CAD模型相似度比较算法。由于模型为边界表示，故可将比较的两个CAD模型用属性邻接图来表示。根据模型面的属性集定义排序函数，将属性邻接图的顶点按照一定的序列排列。同时定义了顶点间的关系集，可将排序后的属性邻接图映射为字符串。通过编辑距离来度量字符串的相似度，并用字符串的相似度来度量CAD模型的形状相似度。实验结果表明，该算法能客观地反映两个CAD模型的相似度，且准确性和效率令人满意。

英文摘要：

This paper presented an algorithm based on string comparison to assess the similarity of CAD models. Boundary representation（B--rep） model was represented by AAG. Permutation function was defined to put the vertexes in AAG in certain order according to attributes of vertexes. Given the relationship between vertexes, the permuted AAG can be mapped into a string. Edit distance was then used to calculate the distance between the two strings and the distance suggested the shape similarity of the two CAD models. The experimental results show that the proposed algorithm can give the degree of similarity of the two CAD models accurately and efficiently.