On fuzzy rule-based algorithms for image segmentation using gray-level histogram analysis

One of the biggest problems in computer vision systems, analyzing images having high uncertainty/vagueness degree, is the treatment of such uncertainty. This problem is even clearest in the segmentation process. Fuzzy set theory and fuzzy logic are ideally suited for dealing with such uncertainty. This work extends our earlier and on-going work in automated image labeled segmentation, modeled following expert's knowledge. This knowledge is represented by means of a fuzzy rule-base, wherein the membership functions associated to the labels are defined based on the analysis of the gray-value's histogram of the pixels of the training images. The proposed system has been evaluated on two very different real data sets.