Online SOM-based level set model for image segmentation

DOI

10.2991/icmmita-16.2016.151How to use a DOI?

Keywords

SOM; Level set model; Local information; Image segmentation.

Abstract

In this paper, a novel region based active model integrating self organizing map (SOM) for image segmentation is proposed. The SOM is utilized to describe image intensity distributions in an unsupervised way and establish the entire weights set of the neurons. Then, the weights are divided into two branch sets depending on whether the weight is larger than the weight mean. Further we obtain the means of two weights subsets, which could be regarded as the average intensities of the background and foreground, respectively. In addition, the global intensities updated during contour evolution are replaced by the prototypes of the entire weights set according to certain rule. Along with the two weight means, the global means are averaged to create the threshold for the whole image to guide the contour movement. The proposed model could be successfully applied for image segmentation. Experiment results demonstrate the effectiveness of our model.

Copyright

© 2017, 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  - Xiaomin Xie
PY  - 2017/01
DA  - 2017/01
TI  - Online SOM-based level set model for image segmentation
BT  - Proceedings of the 2016 4th International Conference on Machinery, Materials and Information Technology Applications
PB  - Atlantis Press
SP  - 813
EP  - 816
SN  - 2352-538X
UR  - https://doi.org/10.2991/icmmita-16.2016.151
DO  - 10.2991/icmmita-16.2016.151
ID  - Xie2017/01
ER  -