Combining colour spaces: a multiple classifier approach to colour texture classification
Samuel Chindaro · 2003
We propose a novel approach to colour texture classification based on combinations of the information included in different colour spaces. Our approach is based on recent advances in features extracted using Gaussian Markov random fields. A number of comparative works on colour spaces have been presented, but not much has been done on combining the colour spaces to produce more robust discrimination systems. The work is an empirical study of decision combination approaches using classifiers obtained through training in various colour spaces and sub-spaces. We include results of experiments carried out using individual and combinations of six different colour spaces and their chromatic sub-spaces. Our results lead to the conclusion that colour texture classification can benefit significantly from techniques based on combining decisions obtained from classifiers trained on different colour spaces and sub-spaces.