3D Assembly Model Retrieval Method Based On Assembly Semantic

From '"made in China" to "created in China",3D assembly model is a basic data source that needs to be attached much importance in the strategic transformation.To ensure the economic efficiency of enterprises,the reusability of these models is key,and the retrieval technology of the 3D assembly model is an effective way to improve the model's reusability.Most existing 3D model retrieval technologies are based on the retrieval of a single part of the model,which is used as the retrieval condition to find similar models from the model library.However,in the new era,the design and production of enterprises are not limited to the retrieval of a single part of the model,and the growing demand is for the intelligence of the model's retrieval technology.When a user takes a single part of a model as an input model,the system can find the parts in the model library that match the input model.Therefore,this paper uses a 3D assembly model retrieval method based on assembly semantic,the specific research contents are as follows:3D assembly model similarity evaluation method based on assembly semantics.First,the assembly semantics are classified according to the assembly design intention,which summarizes the assembly relations among the assembly models.Furthermore,ontology technology is adopted to express assembly semantics explicitly,so that assembly semantics can be stored in the computer in a certain form.Then,based on the empirical calculation method and combining with the characteristics of assembly semantics,the quantitative and qualitative calculation of assembly semantics are carried out respectively to obtain the calculated value of assembly semantic similarity.Finally,based on the bipartite graph matching method and kuhn-munkres algorithm,the similarity evaluation of model assembly semantics in the 3D model library is carried out,and the 3D model that conforms to the assembly semantics of input model is extracted.3D model retrieval based on semantic reduces the amount of data for subsequent 3D model geometrical retrieval and improves the efficiency of the whole retrieval method.3D assembly model retrieval method based on graph isomorphism.In order to achieve the goal of precise retrieval,and retrieve the model which can be assembled with the input model,the geometric interfaces with assembly relations are compared.Firstly,the 3D model is represented by the attribute adjacency graph for similarity comparison.Then,there is a conjugate relationship between the mating parts of the assembly parts,that is,the topological connection of the mating parts is the same but the concavity and convexity are opposite,so the conjugate subgraph exists in the attribute adjacency graph between the models with assembly relations.Finally,on the basis of graph isomorphism judgment,frequent subgraph mining method is adopted to extract the attribute adjacency graph satisfying conjugate subgraph,so as to retrieve the required 3D model.In the case of multi-assembly interface,it finds the atlas containing the same attribute information and topology relation as the input model from the atlas library.It serializes attribute adjacency graph and reduces the number of vertexes in graph isomorphism.On the basic of frequent subgraph mining,it makes cyclic judgment on the atlas in the atlas library,and extracts the 3D assembly model with multiple corresponding conjugate interfaces that can be assembled with the input model.In order to realize the retrieval method mentioned above,3D assembly model retrieval system based on assembly semantic is constructed.Based on the general 3D model retrieval system framework,the requirements of the retrieval system are analyzed,the functional modules of the retrieval system are constructed,and the architecture of the retrieval system is designed.Taking the 3D machine tool model accumulated by the research group as the experimental object,the 3D assembly model retrieval of single assembly interface and multiple assembly interfaces was experimentally verified from the semantic and geometric perspectives.