Use of fuzzy clustering for determining mass functions Dempster-Shafer theory

This paper presents a new approach of automatically determining mass functions in Dempster-Shafer (evidence) theory from the grey levels of images. To achieve this, two underlying aspects have been in particular investigated in the context of segmentation of images from two different sources. The first one concerns the determination of mass function using the fuzzy C-means clustering, and the second one deals with the matching of clusters in the two images being fused. The proposed approach is illustrated with the aid of simulations and examples on physical images