Research On Retrieval Method Of 3d Relief Model Based On Ontology Semantic Analysis

With the development of computer graphics and 3D printing technology,digital relief artwork editing has been gradually become one of the most heated topics in computer graphics research.Reliefs usually assemble a scene with several 3D models which have internal relationship.Therefore,to effectively and accurately retrieve these 3D models from the huge 3D model library is the primary problem for digital relief creation.There are some problems in existing 3D model retrieval methods,such as low retrieval efficiency and low precision.The methods of 3D model retrieval in relief are desperately need a solution to find semantic relevance of retrieval methods.This paper proposes a retrieval method of 3D relief models based on ontology semantic analysis,and builds a retrieval system of 3D relief model to realize this method.The experimental results show that the system can retrieve the models in 3D relief model database and return results with satisfaction accuracy and effective.The main research contents and conclusions are as follows:(1)Construction of Semantic Web of 3D Relief ModelTo solve the problem of the lack of research and construction of ontology semantic web suitable for relief field,this paper constructs 3D relief model ontology and semantic web based on "seven-step method" and Baidu Encyclopedia semantic relationship extraction.The method is divided into three steps: first,the relief models in the 3D model base are classified,and then the ontology model is established based on the classification results;secondly,the semantic association is established according to the correlation between the models in the relief scene;finally,a rule-based semantic reasoning framework is constructed to solve the problems of incompleteness and error in establishing the association,and the automatic semantic relationship between the models is realized.(2)Semantic Similarity Computation Based on Ontology Tree Structure and SynonymThis paper proposes an improved semantic similarity calculation method based on ontology tree structure and synonym,which was expected to address the problems of low recall and precision in 3D model retrieval.This method takes Synonym Words Forest compiled by HIT-SCIR as the basic knowledge data,and calculates the similarity between ontology vocabulary and users' input to determine the similarity between them.This method can not only be used to add 3D models to the model database,but also can calculate the similarity between retrieval items and model base labels.The experimental results show that the recall and precision of the retrieval method with similarity calculation are improved by 21.9% and 7.8% respectively compared with the traditional keyword matching-based retrieval method.(3)Design and Implementation of 3D Model Retrieval SystemBased on the above-mentioned three-dimensional model semantic web and semantic similarity calculation,this paper designs and implements a 3D relief model retrieval system.The system integrates SPARQL language with Jane API to realize back-stage logic,and develops friendly interface by Qt platform.The system also implements a feedback mechanism to retrieve the users' choice again.