COLOR IMAGE SEGMENTATION BY FUZZY C-MEANS CLUSTERING ALGORITHM

Abstract

The segmentation of colors from complicated images has since long been a continuous area of research due to the practical significance of it’s results in various applications especially the medical imaging field. Clustering of images is a well known technique of segmenting images. We accomplish fuzzy clustering by including fuzzy sets in the results. The most popular Fuzzy c-means clustering algorithm (FCM) often used for segmenting images, has been studied and applied to the test images, and its pros and cons weighed. As an attractive extension of the FCM algorithm, we introduce the concept of co-clustering, which simultaneously clusters objects and features, producing different memberships for both. An important existing co-clustering algorithm developed for document clustering called the “Fuzzy co-clustering algorithm for categorical and multivariate data” (FCCM), is applied to various test images. The defects of the FCCM algorithm when applied directly to images is studied in detai...