Image segmentation has been, and still is, a relevant research area in Computer Vision, and hundreds of segmentation algorithms have been proposed in the last 30 years. Many segmentation methods are based on two basic properties of the pixels in relation to their local neighbourhood: discontinuity and similarity. Methods based on pixel discontinuity are called boundary-based methods, whereas methods based on pixel similarity are called region-based methods. However, it is well known that such segmentation techniques - based on boundary or region information alone - often fail to produce accurate segmentation results . Hence, in the last few years, there has been a tendency towards algorithms which take advantage of the complementary nature of such information.
Reviewing the different works on region-based segmentation which have been proposed (see surveys on image segmentation [2,3]), it is interesting to note the evolution of region-based segmentation methods, which were initially focused on grey-level images, and which gradually incorporated colour, and more recently, texture. In fact, colour and texture are fundamental features in defining visual perception and experiments have demonstrated that the inclusion of colour can increase the texture segmentation/classification results without significantly complicating the feature extraction algorithms . Nevertheless, most of the literature deals with segmentation based on either colour or texture, and there is a limited number of systems which consider both properties together.
In this work we propose a new strategy for the segmentation of colour texture images. Having reviewed and analysed more than 50 region-boundary cooperative algorithms, we have clearly identified 7 different strategies (see ) to perform the integration. As a natural development of this review work, we defined a new strategy for image segmentation  based on a combination of different methods used to integrate region and boundary information. Moreover, to knowledge of the authors, there has not yet been any proposal which integrates region and boundary information sources while taking colour and texture properties into account. Hence, we have extended our previous approach to deal with the problem of colour texture segmentation. We focus on ``color texture'' taking into account that is both spatial and statistical. It is spatial since texture is the relationship of groups of pixels. Nothing can be learned about texture from an isolated pixel, and little from a histogram of pixel values.
The remainder of this paper is organised as follows: a review of the recent work on color texture segmentation concludes this introduction. Section 2 describes the proposed segmentation strategy detailing the placement of starting seeds, the definition of region and boundary information and the growing of active regions. The experimental results concerning a set of synthetic and real images demonstrating the validity of our proposal appear in Section 3. Finally, conclusions are given in Section 4.