A Novel Approach Based on Hybridization of Fuzzy C-MEANS and Competitive Agglomeration for Image Segmentation

Abstract

Image segmentation is the process of partitioning a digital image into multiple segments. Different methodologies have been proposed for the segmentation based on normal techniques such as region growing, threshold technique, watershed transform. The disadvantage of these methods leads to the development of segmentation based on clustering techniques. The main concept of data clustering is to use the centroid to represent each cluster. Also it is based on the similarity between the input vectors to that of the centroid to represent each cluster. Parametric and Non-parametric methods are the broad classes of the clustering methods. Non-parametric method involves finding natural groupings in a dataset using a Euclidean distance between the samples of the dataset. Non-parametric clustering includes k-means, hierarchical, spectral clustering. The disadvantages of these methods are lack of sufficient robustness to image noise. So, a fuzzy segmentation methodology has been widely applied in image clustering and segmentation. The important problem in fuzzy c-means is to specify the number of clusters and selection of objective function. So, a fuzzy clustering-based vector quantization algorithm is used. This algorithm utilizes a specialized objective function, which involves the fuzzy c-means along with a competitive agglomeration term. This algorithm is a fast process and the reconstructed images maintain high quality.