Adaptive multidimensional fuzzy sets for texture modeling

Abstract

The modeling of the perceptual properties of texture plays a fundamental role in tasks where some interaction with subjects is needed. In order to face the imprecision related to these properties, fuzzy sets defined on the domain of computational measures of the corresponding property are usually employed. In this sense, the most interesting approaches show that the combination of different measures as reference sets improve the texture characterization. However, the main drawback of these proposals is that they do not take into account the subjectivity associated with human perception. For example, the perception of a texture property may change depending on the user, and in addition, the image context may influence the global perception of a given property. In this paper, we propose to solve these problems by combining the use of several computational measures in a reference set with adaptation to the subjectivity of human perception. To do this, we propose a generic methodology that automatically transforms any multidimensional fuzzy set modeling a texture property to the particular perception of a new user or to the image context. For this purpose, the information given by the user, or extracted from the textures present in the image, are employed.