A Study of Various Clustering Algorithms for Image Segmentation

The major task of computer vision applications is discriminating the objects of interest. In a particular image in order to identify the objects, we segment the image into homogeneous regions such that an image is illustrated into something that is simple to state and easy to study so that selecting the objects of interest is also an easy task. This entire process is known as image segmentation. Segmentation plays a vital role in illustrating and analyzing the satellite images and it is a crucial section of image exploration system. To segment any image, distinct techniques such as clustering, thresholding, watershed, neural networks, etc. are used. One method that is frequently used to deal with the problem of many of an image’s pixels staying identifiable is clustering. But the segmentation technique alone cannot achieve better segmented results. In the process of image segmentation to improve the efficiency of segmented results, the segmentation technique is combined with either optimization algorithm or entropy measures or ensemble of different algorithms is used. In this study, we present different kinds of optimization algorithms, combinations of clustering algorithms and their advantages and also various categories of existing clustering based techniques used for image segmentation as well as pros and cons of each technique.