Learning prototypes for building fuzzy color spaces

In this paper, an approach to learn fuzzy color spaces from images is proposed. Fuzzy colors were introduced as a suitable way to model color categories, being the newest and best performance approaches those which use crisp color prototypes in order to build fuzzy color spaces. Current proposals in literature use standard set of colors prototypes, like ISCC-NBS ones, but it is shown that they are not flexible enough to collect the particular colors of a given image. To address this problem, in this paper we propose to extract image relevant colors and use them as prototypes to model fuzzy color spaces. We illustrate the advantages of our approach on the basis of some real experiments.