Title: A fast fuzzy c-means algorithm for colour image segmentation

Authors: Qigang Liu; Li Zhou; Xiangyang Sun

Addresses: Sydney Institute of Language and Commerce, ShangHai University, ShangHai 201800, China ' Sydney Institute of Language and Commerce, ShangHai University, ShangHai 201800, China ' Sydney Institute of Language and Commerce, ShangHai University, ShangHai 201800, China

Abstract: Image segmentation as a big part of computer vision has been the hot research topic nowadays. Fuzzy c-means (FCM) algorithm, as the standard and basic approach, is widely adopted by most researchers. In this paper, authors deeply analyse some defects of this technique, such as presetting clustering centre, initialling cluster centres randomly and heavy computation load. To overcome these drawbacks, a novel approach is proposed based on the combination of hierarchal subtractive clustering and fuzzy c-means, in which hierarchal subtractive clustering is introduced for simplifying the cluster centres fixing process and speeding up convergence. A series steps has been offered to imply how to implement this approach, and finally, experiments on two colour images show that, without damaging clustering validity and effectiveness, the novel approach is superior on clustering speed compared to FCM, it is the better choice for dealing with today's high resolution colour images.

Keywords: fuzzy c-means; FCM; hierarchal subtractive clustering; membership matrix; density; colour images; image segmentation; computer vision.

DOI: 10.1504/IJICT.2013.054939

International Journal of Information and Communication Technology, 2013 Vol.5 No.3/4, pp.263 - 271

Received: 13 Dec 2012
Accepted: 06 Jan 2013

Published online: 19 Dec 2013 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article