K-Means Cluster Analysis for Image Segmentation
S. M.AqilBurney, Humera Tariq · International Journal of Computer Applications · 2014
Does K-Means reasonably divides the data into k groups is an important question that arises when one works on Image Segmentation?Which color space one should choose and how to ascertain that the k we determine is valid?The purpose of this study was to explore the answers to aforementioned questions.We perform K-Means on a number of 2-cluster, 3cluster and k-cluster color images (k>3) in RGB and L*a*b* feature space.Ground truth (GT) images have been used to accomplish validation task.Silhouette analysis supports the peaks for given k-cluster image.Model accuracy in RGB space falls between 30% and 55% while in L*a*b* color space it ranges from 30% to 65%.Though few images used, but experimentation proves that K-Means significantly segment images much better in L*a*b* color space as compared to RGB feature space.