Segmentation of Images using Automatic Fuzzy Clustering Framework

Clustering is grouping up of data points. Using clustering algorithms, the data points can be grouped with similar properties. Fuzzy clustering is grouping of data points of clusters of one or more clusters. Density Peak (DP) clustering can find the clusters but when the sum of clusters is increased, it suffers memory overflow, because when a normal size image is used for image segmentation which contains a greater number of pixels, it results in a high degree of similarity matrix. To avoid this, Automatic Fuzzy Clustering Framework (AFCF) for segmentation of image could be introduced. This framework contributes in three ways. To begin with, the Density Peak method is used for the concept of Super Pixel, which decrements the similarity matrix size and there by enhances DP algorithm. Secondly, the Density Balance technique generates a stable decision graph, which enables the DP algorithm for a fully autonomous clustering. Lastly, to improve image segmentation outcomes, the system which works on prior entropy employs a Fuzzy c-means clustering. Through this, information of pixels in spatial neighbors are considered and can see improved segmentation results. In the present work, an attempt is made to develop and explain the segmentation of images using Automatic Fuzzy Clustering Framework.