An imprecision-based image classifier
P. Malinski, Sandra Aparecida Sandri, C. Freitas · 2005
Proposes an image classification procedure that initially performs an imprecise classification of the image, by assigning a set of classes to each pixel, and then refines this classification interactively, by verifying each pixels neighbours classification, until all the pixels in the image are precisely classified. We also propose an implementation of this approach, based on possibility theory and genetic algorithms, and present some results for SAR images.