Authors:
Jun-Yang Lin
1
;
May-Fang She
1
;
Ming-Han Tsai
1
;
I-Chen Lin
1
;
Yo-Chung Lau
2
and
Hsu-Hang Liu
2
Affiliations:
1
National Chiao Tung University, Taiwan
;
2
Telecommunication Laboratories, Chunghwa Telecom Co. and Ltd, Taiwan
Keyword(s):
3D Object Retrieval, Depth Image Analysis, Shape Matching.
Related
Ontology
Subjects/Areas/Topics:
Advanced User Interfaces
;
Applications
;
Computer Vision, Visualization and Computer Graphics
;
Geometry and Modeling
;
Image-Based Modeling
;
Interactive Environments
;
Pattern Recognition
;
Software Engineering
Abstract:
Existing 3D model retrieval approaches usually implicitly assume that the target models are rigid-body. When
they are applied to retrieving articulated models, the retrieved results are substantially influenced by the model
postures. This paper presents a novel approach to retrieve 3D models from a database based on one or few
input depth images. While related methods compared the inputs with whole shapes of 3D model projections
at certain viewpoints, the proposed method extracts the limbs and torso regions from projections and analyzes
the features of local regions. The use of both global and local features can alleviate the disturbance of model
postures in model retrieval. Therefore, the system can retrieve models of an identical category but in different
postures. Our experiments demonstrate that this approach can efficiently retrieve relevant models within a
second, and it provides higher retrieval accuracy than those of compared methods for rigid 3D models or
models wit
h articulated limbs.
(More)