Hits:

Indexed by:期刊论文

Date of Publication:2012-01-01

Journal:ICIC Express Letters

Included Journals:EI、Scopus

Volume:6

Issue:10

Page Number:2679-2684

ISSN No.:1881803X

Abstract:The fuzzy c-means algorithm (FCM) is a well-accustomed tool for image segmentation. However, as most of the clustering algorithms, FCM cannot segment images with damaged blocks directly. Therefore, a method for segmenting images with damaged blocks is proposed in this paper, which is based on the classical OCS-FCM algorithm. Firstly, gray value and several neighborhood features are employed to construct the image feature space, so that the clustering algorithms for incomplete dataset can be used to segment damaged images. Then, a novel segmentation algorithm based on OCS-FCM is proposed, in which a simplified image energy representation is added to the clustering objective function. Experimental results demonstrate that the proposed algorithm achieves better segmentation performance than the compared algorithms. ? 2012 ISSN.