3D Model Retrieval Based on a Dynamic Multi-Descriptor Fusion Algorithm

• Main Authors: Yao-Wen Hou, 侯堯文
• Other Authors: Jau-Ling Shih
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/04832267353230205139

碩士 === 中華大學 === 資訊工程學系(所) === 99 === In this paper, we will propose a dynamic multi-descriptor fusion (DMDF) approach to improving the retrieval accuracy of content-based multimedia retrieval systems. First, an independent retrieval list is generated for each individual descriptor. Second, the automatic relevant/irrelevant models selection (ARMS) approach is proposed to automatically select the relevant and irrelevant models in a 3D model retrieval system without any user interaction. A weighted distance measure, in which the weight associated with each individual descriptor is learnt automatically according to the selected relevant and irrelevant models, is used to measure the distance between two 3D models. Furthermore, a descriptor-dependent adaptive query point movement approach is employed to update every feature vector. This set of new feature vectors will be used to index 3D models in the next search process. Four 3D model databases were used to compare the retrieval performance of the proposed DMDF approach with many state-of-the-art approaches. Experiment results have shown that our proposed DMDF approach provides a promising retrieval result and always yields the best retrieval accuracy.