Rough-Fuzzy Local Information C-Mean Clustering and Collaborative Fuzzy Local Information C-Mean Clustering

In this paper we hybridize fuzzy local information c-mean (FLICM) algorithm with rough c-mean algorithm (RCM) for image clustering. In addition to all advantages of usual FLICM, this new algorithm has a higher convergence rate and better results. In the proposed algorithm for improving the rate of convergence in the primal iterations, we consider a new and different initialization. The algorithm starts with initializing the centroids (cluster centers) vector instead of membership matrix.Also collaboration in FLICM algorithm for image clustering was inspected. This variant of algorithm leads to interesting results. However the speed of algorithm decreased, but from the application point of view it can be useful. To improving the speed convergence, we used particle swarm optimization (PSO) algorithm to consider suitable initial values for each subpopulation.

کلیدواژه ها:

Rough set ، image clustering ، collaborative clustering ، image segmentation ، fuzzy c-mean ، local information ، particle swarm optimization