journal6 ›› 2015, Vol. 36 ›› Issue (2): 29-32.DOI: 10.3969/j.issn.1007-2985.2015.02.007

• Computer • Previous Articles     Next Articles

Improved FCM Image Segmentation Based on Multi-Resolution Analysis and K-Means Clustering

 GUO  Hai-Tao, ZHAO  Hong-Ye, XU  Lei, HOU  Yi-Min, JIAO  Sheng-Xi   

  1.  (1.College of Electronic Information Engineering,Inner Mongolia University,Hohhot 010021,China;2.College of Electrical Engineering,Northeast Dianli University,Jilin 132012,Jilin China;3.College of Automation Engineering,Northeast Dianli University,Jilin 132012,Jilin China)
  • Online:2015-03-25 Published:2015-04-28

Abstract: The fuzzy C-Means (FCM) clustering is widely used in image segmentation,but the random determination of initial clustering centers of the FCM clustering is likely to generate incorrect segmentation of an image.To avoid the such deficiency,a method of choosing initial clustering centers in the FCM clustering for image segmentation is proposed.The method determines the number for the image clustering by means of the number of the peaks in the two-dimensional histogram of an image comprised of gray values of pixels and mean values of their neighborhoods.Then the K-means clustering is used to obtain the initial clustering centers of the FCM clustering for the low-frequency subband image of the original image.The image segmentation experiments show that the proposed method is feasible.

Key words: two-dimensional histogram, multi-resolution analysis, K-means clustering, FCM clustering, image segmentation

WeChat e-book chaoxing Mobile QQ