ISSN: 1546-962X
Published Online: 8 May 2007
Page Count: 13

Utilization of Dempster-Shafer Theory of Evidence in Unsupervised Image Segmentation

Zribi, M.
Université du Littoral Côte d’Opale,

Rekik, A.
Université du Littoral Côte d’Opale,

Benjelloun, M.
Université du Littoral Côte d’Opale,

(Received 5 July 2006; accepted 19 March 2007)

Abstract

In this paper, we propose a new method based on the Dempster-Shafer theory of evidence for unsupervised image segmentation. This theory turns out to be quite efficient in classification of multisensor information. The application of the evidence theory in fusing information coming from different sources still poses certain problems. Of paramount importance is the problem of estimating the belief functions. Due to the coherence of this theory with the Bayesian approach, a parametric algorithm to estimate these functions based on the maximum likelihood method can be realized to estimate these belief functions. The proposed method is validated by experiments on both synthetic and real images. The experimental results show the interest of the algorithm and its potential.

Keywords:
evidence theory, belief functions, fusing information, Bayesian approach, unsupervised segmentation

Paper ID: JAI100658
DOI: 10.1520/JAI100658
ASTM International is a member of CrossRef.