Pixel classification through divergence-based integration of texture methods with conflict resolution

Domènec Puig and Miguel Angel García

domenec.puig@urv.cat, miguelangel.garcia@uam.es

Abstract

This paper presents a n w teehnique for combining multiple texture feaoure extraction methods in order tt classify the pixels of an input image into a Ret of texture msdels of interest. The roblem of integrating multiple texture methods for classifncation purposes is cast as a collaborative decision making problem. Each texture-method is considered to be an expert that gives an openion about the membership of every input image pixel toeeach texture model, along with a conviction about that judgement. A conviction measure based on the Kullback J-divergence between texture models is proposed, along with an arbitration mechanism that combines those convictions by taking into account conflicts that9h2y occur when different experts disagree with a similar strength. The proposed technique is compared to previous pixel-based texture classifiers by using real textured images.