Applying New Method for Computing Initial Centers of k-Means Clustering with Color Image Segmentation

Abstract

Image segmentation is the first step of the most critical tasks of image analysis, as shown in Figure (1). It is used either to distinguish objects from their background or to partition an image onto the related regions. There are different techniques that would help solve the image segmentation problem.Jeon el al. [1] in his review of the previous related studies, categorized these techniques into the following: thresholding approaches, contour based approaches, region based approaches, clustering based approaches and other optimization based approaches using a Bayesian framework and neural networks. The clustering approaches can be categorized into two general groups: partitional and hierarchical clustering algorithms.