Automatic Face Segmentation Based on the Level Set Method

DOI

10.2991/citcs.2012.259How to use a DOI?

Keywords

face segmentation; level set method; signed pressure force function; GAC model; C-V model

Abstract

Face segmentation is a crucial problem in face recognition and other areas like pedestrian detection. Level set methods have been widely used in image segmentation. In this paper, the modified level set evolution equation is used in the level set method to segment the face. Based on the signed pressure force (SPF) function, a variable parameter signed pressure force (VPSPF) function is proposed. The proposed function avoids the disadvantage of the SPF function because it does not need the image intensity to satisfy specific condition. Our method combines the GAC model and the C-V model. Let the VPSPF function which contains the statistical information inside and outside the contour substitute the edge stopping function (ESF) in the GAC model. It not only has the advantages which the Selective Binary and Gaussian Filtering Regularized Level Set (SBGFRLS) method demonstrates but also can solve the problem the SBGFRLS method cannot handle. Experiment on the synthetic image using our method shows better performances than using the SBGFRLS method. Experiment results in the images from the ORL Database of Faces and the general images which have more than one faces show our method's validity and superiority

Copyright

© 2012, 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  - CONF
AU  - Chaoyi Zhang
AU  - Yanning Zhang
AU  - Zenggang Lin
PY  - 2012/11
DA  - 2012/11
TI  - Automatic Face Segmentation Based on the Level Set Method
BT  - Proceedings of the 2012 National Conference on Information Technology and Computer Science
PB  - Atlantis Press
SP  - 1021
EP  - 1024
SN  - 1951-6851
UR  - https://doi.org/10.2991/citcs.2012.259
DO  - 10.2991/citcs.2012.259
ID  - Zhang2012/11
ER  -