Automatic determination of mass functions in Dempster-Shafer theory using fuzzy c-means and spatial neighborhood information for image segmentation

Abstract

The Dempster-Shafer (DS) evidence theory is a new approach to the problem of segmenting multimodal images coming from different sources. The performance of such segmentation scheme is, however, largely conditioned by the appropriate determination of mass functions in DS evidence theory. We present a method of automatically determining the mass function for image segmentation problems. The idea is to link, at the image pixel level, the notion of mass functions to that of membership functions in fuzzy logic. The mass assigned to a pixel is obtained from both the membership degree of the current pixel and those of its neighboring pixels. The membership degree of each pixel is determined by applying fuzzy c-means (FCM) clustering to the gray levels of the image. A method is presented to determine the simple or composite classes in DS evidence theory from the obtained membership degree. Final segmentation is achieved using the DS combination rule and decision. The developed mass function determination method is illustrated with both simulations and examples of physical images. We demonstrate the value of introducing fuzzy clustering in evidence theory for image segmentation.