An accelerated clustering algorithm for segmentation of grayscale images

Conventional clustering techniques like FCM, K-Means, Mountain clustering etc. face the main problem of excessive data while dealing with the very big size images. Due to higher order dependency of clustering techniques on the number of data points, time complexity increases excessively while dealing with very large size images. This paper proposes an advanced version of mountain clustering technique, Fast Mountain clustering (FMC), for segmentation of grayscale images whose run time is almost independent of size of image. The proposed approach consists of defining the dataset in another domain which makes the clustering almost independent of size of the data. The obtained results are compared with the widely used techniques like FCM, K-Means, IMC and found out to be better on the basis of cluster validity measures Global silhouette index (GS) and Partition Index (SC).