Possibilistic Concepts Applied to Soft Pattern Classification

This chapter presents three applications of possibilistic concepts in the domain of soft pattern classification. The first one is on pixel-based image classification where an approach referred as Iterative Refinement of Possibility Distributions by Learning (IRPDL) is presented. The IRPDL approach is based on the use of possibilistic reasoning concepts exploiting expert knowledge sources as well as “ground” possibilistic seeds learning. The second application is on spatial unmixing based on possibilistic similarity. The approach presented here exploits possibilistic concepts to provide flexibility for the integration of both contextual information and a priori knowledge. Possibility distributions are first obtained using a priori knowledge given in the form of learning areas delimitated by an expert. These areas serve for the estimation of the probability density functions of different thematic classes also called endmembers. Finally, the third application is about image segmentation based on possibilistic concepts. The aim of a segmentation process is to distinguish homogeneous regions within an image that belong to different objects. The image segmentation approach is based upon the use of possibility theory concepts and imitating human reasoning scheme when analyzing an observed image in terms of its constituting homogeneous regions.