In this paper, a new algorithm for similar 3D CAD model difference examination based on geometric matching is presented. Firstly, using the boundary representation (B-rep) method, the two 3D models are decomposed into two sets of surfaces, each with an attributed adjacency graph (AAG) which is established using adjacency relationship of corresponding surfaces. The vertices of the AAG are set as the geometric information about surfaces (i.e. surface type, area). The edges of the AAG present the adjacency between surfaces, and the attribute information (i.e. the type and length of edges, the angle between two adjacent surfaces) is also stored in the AAG. Secondly, the surface similarity between two models is calculated according to their types, areas, composition edges and topological relationships. At the same time, the similarity matrix which stores the surface similarity coefficients is generated to find the geometric and topological optimal matching surfaces. Then, in the AAG, with the corresponding vertices of the optimal matching surface pair as the center, the remaining surfaces of two models are quickly and optimally matched according to the topological connections and similarity coefficients while the unmatchable ones are defined as added or deleted surfaces. Finally, differences between the two models are evaluated by analyzing and comparing the geometric attribute information about the matched surfaces.
In order to validate the effectiveness and feasibility of the proposed algorithm, a software prototype for similar model difference examination has been developed. The effectiveness and feasibility of the algorithm have been verified by engineering applications through the industrial needs. The results show that this algorithm can effectively compare the differences among different design iterations and demonstrate its potentials for a wide range of engineering design iterations examination problems.