This paper focusses on the segmentation of colour texture images. The foremost purpose of this approach is to investigate the role of colour information in the analysis of textures. Discrete Cosine Transform is a ltering approach used to extract features from the luminance plane. Each lter mask captures a particular property of the local texture structure. The distribution of chromaticity features with lter masks are used for colour texture description. An unsupervised texture segmentation method was adopted to achieve the segmented result. The performance of the method was evaluated using ROC curve analysis. The results species the importance of the incorporation of colour