Shape Degeneration and Multi-object Search Basing on Multi-shape Priors

DOI

10.2991/ijcis.2011.4.4.6How to use a DOI?

Keywords

image segmentation, level set, shape priors, multi-object segmentation

Abstract

This paper proposed a framework of segmenting multiple targets basing on multiple shape priors. The key novel idea in the proposed framework is the shape degeneration model. By breaking the balance between the competition regions adaptively, the proposed model can degenerate the region which is with low similarity with any of the shape priors, until this region completely disappears. Meanwhile the proposed model can also merge the regions that belong to a single object automatically, which is especially important when some of objects are in partial occlusion. Additionally, we presented two new models in the proposed framework. One is a new local model of active contour, which can segment image with intensity inhomogeneity correctly. The other is a generalized shape prior model, which extends the intrinsic alignment model, and can take extra affine parameters into account in the alignment process of shape priors. Experimental results show the validity of the proposed method.

Copyright

© 2011, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - JOUR
AU  - Song Chunhe
AU  - Zhao Hai
AU  - Jing Wei
AU  - Zhu Hongbo
PY  - 2011
DA  - 2011/06/01
TI  - Shape Degeneration and Multi-object Search Basing on Multi-shape Priors
JO  - International Journal of Computational Intelligence Systems
SP  - 467
EP  - 475
VL  - 4
IS  - 4
SN  - 1875-6883
UR  - https://doi.org/10.2991/ijcis.2011.4.4.6
DO  - 10.2991/ijcis.2011.4.4.6
ID  - Chunhe2011
ER  -