Automated Two-Dimensional Fuzzy C-means with K-means clustering for Image Segmentation

Segmentation is a technique that divides an image into many regions that have strong relations with objects to demonstrate the real information collected from the real world. We presented hybrid method of A2DFCM and K-means clustering algorithm for image segmentation. This paper introduces the Automated Two-Dimensional Fuzzy C-mean with K-means (A2DFCM-KM) algorithm. It is a novel unsupervised clustering technique. The proposed technique differs from the conventional clustering technique because user needs not to be determining the number of clusters. A2DFCM-KM relates to local and spatial information of the data into the clustering analysis. By this proposed method we minimized CPU time to about one sixth compare to other existing methods without affecting the overall quality of terminal clusters produced. During iteration we will reduce an objective function that shows the distance between data points and cluster center weights using membership value. Simulation results show that the proposed method outperforms the existing A2DKM method qualitatively as well as quantitatively as it is less sensitive to noise and also eliminates noisy spots. The results also shows that the proposed method reduces false blobs and also more homogeneous regions are obtained by it.