Researches On 3d Model Retrieval Based On XML Technology And Semantic Matrix

With the rapid development of Internet and information technology, 3D models have become the fourth multimedia data types following sounds, images and video. The technology of 3D model retrieval became the research focus in multimedia information retrieval. The research on 3D model retrieval exist deficiencies that 3D model descriptive method are not unified and 3D model retrieval exist deficiencies in semantic-based search and 3D models feature combination. These affect the effectiveness and efficiency of 3D model retrieval.In order to solve these problems, this dissertation proposes a 3D model retrieval method combined with XML technology based on researching 3D model retrieval, and continued to study 3D model retrieval based on semantic matrix feedback and feature combination. The main contents of this paper as following:1. A various algorithms of 3D model feature extraction are introduced according to categories. For example, shape-based feature extraction algorithms, visual-based feature extraction algorithms and based on the surface properties feature extraction algorithms. The classic algorithms of each category are introduced in detail.2. At first, a content-based retrieval method is proposed by using XML to describe 3D models. Then an algorithm to improve the differentiation of the models is presented which based on shape feature extraction. Finally, the experiment results show that the XML technology used in 3D model retrieval is feasible, shape-based feature extraction algorithm is more effective than traditional shape-based feature extraction method in 3D model retrieval.3. An integrated method of 3D model retrieval based on semantic matrix feedback and feature combination is proposed. When the user is not clear whether the models are similar or not similar in the relevance feedback, semantic matrix can be used to determine whether the models are similar or not. Further more, multi-feature combination method is combined with semantic matrix feedback in the 3D model retrieval. The experiment results show that this method is more accurate than the single feature extraction method.