A Method of Automatic Semantic Annotation of 3D Model Based on Content Feature

Abstract

In response to the existing problem of 3D model semantic retrieval, a method of automatic semantic annotation of 3D model based on content feature is proposed on the basis of Word Net. According to the similarity of content features, this method selects the annotation vocabularies to construct a vocabulary set. Then, some appropriate vocabularies can be selected from the vocabulary set to annotate the 3D models by the similarity between the 3D model to be annotated and the vocabularies in the vocabulary set. In the experiments, three relative parameters are optimized to improve the performance and efficiency. Proved by the experiments, this method proposed in this paper can solve the semantic gap problem. The performance and efficiency of the method is pretty good.