Similarity Assessment of Assemblies Based on Random Walks and Optimal Matching

In order to improve the utilization of topology information in assembly model retrieval, an efficient similarity analysis method based on the random walks and optimal matching is presented in this paper.Firstly, the parts of assembly are represented as graph nodes, and the assembly is represented by the attribute adjacent graph which is simplified by removing fastener.Secondly, the probability matrix is built based on the connection relationships analysis, and the topological signature of part is computed by using the random walks model.Thirdly, the bipartite graph for assembly is built by the multiple comparisons of topological and attribute signature, and the similarity of assemblies is assessed by solving bipartite graph optimal matching problem.Finally, the validity and rationality of the method is demonstrated with the study cases.