一种基于视图分类的3维模型检索方法  

A 3D Model Retrieval Approach Based View Classification

作　　者：万丽莉[1] 赵沁平[1] 

机构地区：[1]北京航空航天大学虚拟现实与可视化新技术教育部重点实验室,北京100083

出　　处：《中国图象图形学报》2007年第3期529-535,共7页Journal of Image and Graphics

基　　金：国家重点基础研究发展规划"973"项目(2002CB312105)

摘　　要：3维模型检索是近年来基于内容检索的研究热点。在基于2维草图检索3维模型的研究中,通常没有结合人的知识。为了提高3维模型检索的精度,提出了一种基于视图分类的3维模型检索方法,它的主要思想是通过视图分类将人对3维模型的认知变换为对2维视图的理解,以便在度量2维草图与视图之间的相似性时能利用人的知识。该方法涉及以下2个问题:视图分类、2维草图与视图之间的相似性度量。实验结果表明,该方法能提高检索精度,因此可应用在2维草图查询方式的3维模型检索中。3D model retrieval has recently become one of active research areas in content-based retrieval. In the research of 3D model retrieval based 2D sketch, people＇s knowledge is not usually adopted. In order to improve the performance of 3D model retrieval, this paper proposes a 3D model retrieval approach based view classification. The main idea is to transfer people＇s cognition for 3D models to the understanding of 2D views by view classification, so that people＇s knowledge can be used when measuring the similarity between 2D sketch and views. Two problems are involved： view classification and similarity measure between 2D sketch and views. Experiments show that the method can improve the precision of retrieval, and thus the proposed approach can be applied to interactive 3D shape retrieval based 2D sketch.

关 键 词：3维模型检索 2维草图 视图 特征加权 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]