作者: Ayellet Tal , Emanuel Zuckerberger
DOI: 10.1007/978-3-540-75274-5_3
关键词:
摘要: This paper examines the application of human vision theories Marr and Biederman to retrieval three-dimensional objects. The key idea is represent an object by attributed graph that consists object’s meaningful components as nodes, where each node fit a basic shape. A system realizes this approach was built tested on database about 400 objects achieves promising results. It shown representation 3D very compact. Moreover, it gives rise algorithm invariant non-rigid transformations does not require normalization.