Research on Semantic-Based 3D Model Retrieval

In this paper, we present on the research of semantic-based 3D model retrieval system and its developed framework. We use RDF annotation to label 3D models for improving retrieval precision. The framework contains three parts: preprocessing, feature extraction and similarity measurement. The parts are recommended to analyze the 3D model. In the preprocessing part, we deploy PCA to make models normalization. In the feature extraction part, shape component and semantic component are used to construct an efficient 3D object retrieval scheme. In the last, similarity measurements, like Euclidean distance and quadratic form distance, compute the similarity distance between models. The developed framework is designed to improve the efficiency of 3D model retrieval system.