CONTENT BASED 3D MODEL RETRIEVAL: A SURVEY

摘要

In recent decades, Computer aided retrieval, analysis of text, image data and graphical data technology are developing rapidly. However, the development in the field of 3D model retrieval is still in its infant stage. Thus, the related technologies, such as preprocess, normalization, extracting descriptor, index, and similarity measure still get a long way to explore. In this paper, we review recent various methods of content based 3D model retrieval, and propose a taxonomy for these methods. We also evaluate several requirements for 3D shape representation: (1) model file requirements, (2) necessity of pose normalization, (3) robustness, (4) sensitivity, (5) efficiency, (6) multiresolution support, (7) ability to perform partial matching, and (8) discrimination abilities. Finally, we discuss the advantages and disadvantages of various kinds of methods, and identify several issues in the research area of content based description and retrieval of 3D model.