Image Segmentation Based on Moments and Color Map

Image segmentation is to identify homogenous regions in the image. Basically, the main role of image segmentation is to cluster pixels into prominent image regions i.e. the regions which are corresponding to individual surfaces, objects or natural parts of the objects. The moment is an invariant feature which has been used for object recognition as a descriptor. Additionally, color map can also be used for clustering pixels in corresponding region. In this paper color map is used as a feature for clustering purpose. Color map is obtained by finding out labels in an image. For each possible window representation of each pixel in an image, statistical moment values are computed. Now feature vector of each pixel is then classified using k-means clustering method. Quantitative analysis is done to show the accuracy of the used feature vector i.e. moments and color-map.