Computer Engineering and Applications ›› 2014, Vol. 50 ›› Issue (9): 163-166.

Previous Articles     Next Articles

Enhanced Fuzzy C-Means clustering method for infrared image segmentation

WEI Yingzi1,2, LI Jingjing1   

  1. 1.School of Information Science and Engineering, Shenyang Ligong University, Shenyang 110159, China
    2.State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang 110016, China
  • Online:2014-05-01 Published:2014-05-14



  1. 1.沈阳理工大学 信息科学与工程学院,沈阳 110159
    2.中国科学院沈阳自动化研究所 机器人学国家重点实验室,沈阳 110016

Abstract: The numbers of clusters and clustering center have a great effect on the selection of image segmentation results. The traditional Fuzzy C-Means algorithm always adopts empirical values as the number of clusters and clustering center. In order to determine the optimal number of clusters, the validity measure variable of Global Silhouette Index is employed. The most important disadvantage of Fuzzy C-Means clustering is that it generally does not give proper clustering center in the first run. A method which is the minimum-maximum distance based on the gray value of histogram to compute the original clustering center is put forward. Experimental results show that the method is effective and efficient.

Key words: infrared image segmentation, Fuzzy C-Means(FCM) algorithm, clustering center, global silhouette index

摘要: 分类数和初始聚类中心的选取对红外图像的分割结果有较大的影响。传统的模糊C均值算法的分类数和聚类中心往往设定为经验值。为获得最佳的分类数,提出采用轮廓指标确定出较理想的分类数。针对传统的模糊C均值聚类算法对初始聚类中心比较敏感的问题,提出了基于直方图灰度值的最小最大距离法来确定初始聚类中心。实验结果表明该方法有效可行。

关键词: 红外图像分割, 模糊C均值算法, 聚类中心, 轮廓指标