3D CAD Model Retrieval Based On Graph Theory

Recently, with the development of3D modeling technology and computernetwork technology, there is a rapid growth in the number of sharing and reusable3D CAD models. So, how to manage and retrieve effectively the existing3DCAD models and how to improve the sharing and reuse of the3D CAD modelresource need to be solved immediately. Based on it,3D CAD Model Retrievalgradually becomes a hot research field in computer graphics field.Due to3D CAD models are often characterized with complicated geometryand topology, how to help the designers quickly and accurately find the reusableobject models from the database containing massive amount of models using therelatively rough query instance is still a big challenge. This topic mainly studiesthe representation of3D CAD models and3D model retrieval technique3D CADmodels.A novel representation of3D CAD models using hierarchical graph isproposed. Hierarchical graph contains the model information of featuredependency graph (FDG) and property adjacency graph (PAG). Hus shape featuredescriptors and topology relationship descriptors are extracted from HG. In thisway, the descriptors extracted easily and accurately. Meanwhile, the multi-leveldescriptors can not only describe the overall shape of models, but also describethe local details, improve search efficiency of three-dimensional CAD model.On this basis, a3D model retrieval method is proposed based on geneticalgorithm (GA), and ant colony optimization (ACO). GA is adopted to obtainsuboptimal solutions. The pheromone of ACO is initialized according to thesuboptimal solutions, and then a further search among the suboptimal solutions isoperated for better solution. And finally, the optimal solutions of the productdesign can be searched. This method can improve the accuracy and efficiency of3D CAD model retrieval.Based on the above researches, a3D model retrieval system HUST-CMRS isdeveloped. Our experimental results show that, the algorithm proposed in this paper can implement multi-mode indexing and satisfy user’s personalized need.