CAD Assembly Retrieval by Searching Kinematic Relation Isomorphic Sub-graph

Effective reuse of existing models, which can shorten the cycle of the product design, can be achieved by assembly retrieval. In this paper, an assembly retrieval approach is proposed by searching kinematic relation isomorphic sub-graph. First, the assembly constraints among parts in the assembly are gotten by CAD software API, and the degrees of freedom of parts are reasoned according to their constraints to generate the kinematic relation adjacency graph. Then, the graph is divided into several sub-graphs to screen assemblies according to the number of matched sub-graphs. And VF2 algorithm is used to search the maximum kinematic relation isomorphic sub-graph between assemblies, so as to obtain the mapping between parts. Subsequently, point sampling is carried out on surfaces for all parts in the assembly. The distances of arbitrary point-pairs are calculated to generate the histograms of part shape distribution. And the histograms of part-pairs in the maximum isomorphic sub-graph are compared to obtain the dissimilarity of part shape. Eventually, the assembly model retrieval is realized by considering the maximum kinematic relation isomorphic sub-graph and part shape information comprehensively. Experiments show that the approach has high accuracy.