A New Approach for Colour Texture Segmentation Based on SRM

Image processing techniques have become increasingly important in a wide variety of applications due to the advent of computer technology. With the increased application of colour images, the colour image segmentation is more and more concerned by researches. There are many image segmentation techniques for segmenting an image but no one can tell which one is the better. Different techniques are used depending upon the domains. Image segmentation has been subject of considerable research activity and segmentation plays an important role in image understanding, image analysis and image processing. This paper presents an approach for segmenting a coloured image in a perceptual meaning. This approach includes two steps first where an input image is colour quantized by using K means clustering algorithm. Next we apply Statistical Region Merging (SRM) algorithm for segmentation. Finally we are comparing segmented results for quantized image and also unquantized image by using metrics like probabilistic random index (PR), global consistency error (GCE), and variation of information (VI).