Enhanced interval type-2 fuzzy c-means algorithm with improved initial center

• An enhanced interval type-2 FCM is proposed to reduce the complexity of IT2FCM.
• A centroid initialization method is proposed to initial cluster centers.
• The process of type-reduction in IT2FCM is optimized.
• Experimental comparisons on random data clustering and image segmentation are given.

Uncertainties are common in the applications like pattern recognition, image processing, etc., while FCM algorithm is widely employed in such applications. However, FCM is not quite efficient to handle the uncertainties well. Interval type-2 fuzzy theory has been incorporated into FCM to improve the ability for handling uncertainties of these algorithms, but the complexity of algorithm will increase accordingly. In this paper an enhanced interval type-2 FCM algorithm is proposed in order to reduce these shortfalls. The initialization of cluster center and the process of type-reduction are optimized in this algorithm, which greatly reduce the calculation time of interval type-2 FCM and accelerate the convergence of the algorithm. Many simulations have been performed on random data clustering and image segmentation to show the validity of our proposed algorithm.