Computer Engineering and Applications ›› 2012, Vol. 48 ›› Issue (10): 183-186.

Previous Articles     Next Articles

Fast image segmentation of weighted fuzzy C-means clustering based on 2-D entropy

SHA Xiuyan1, WANG Zhenjian2   

  1. 1.School of Mathematics and Information, Ludong University, Yantai, Shandong 264025, China
    2.Information Technology Department, Ludong University Library, Yantai, Shandong 264025, China
  • Online:2012-04-01 Published:2012-04-11

基于快速二维熵的加权模糊C均值聚类图像分割

沙秀艳1,王贞俭2   

  1. 1.鲁东大学 数学与信息学院,山东 烟台 264025
    2.鲁东大学 图书馆情报技术部,山东 烟台 264025

Abstract: In this paper, an image segmentation algorithm for combining fast two-dimensional entropy and weighted fuzzy C-means(FCM) clustering is proposed. The centers of the object and the background are obtained by applying fast two-dimensions entropy algorithm. Then, the influence of every sample on the classification is characterized by the gray difference between the sample and its neighborhood samples. At last, the segmentation is obtained by weighted fuzzy C-means clustering algorithm. The new algorithm can solve the question that the traditional FCM clustering algorithm is sensitive to the initial value. Moreover, it can overcome the shortage that the traditional clustering algorithm is equally partition to the sample set. The experimental result shows that the algorithm not only has good convergence, but also can effectively segment the target from its background. The new algorithm has the important practical application value.

Key words: fuzzy C-means clustering, two-dimensional entropy, image segmentation

摘要: 提出了一种结合快速二维熵和加权模糊C均值聚类的图像分割方法。采用快速二维熵算法对实际图像进行初步分割求得目标和背景的中心,然后采用样本点像素与其邻域灰度像素的差别表征该样本点对分类的影响程度,最后利用加权模糊C均值聚类算法完成图像分割。该方法一方面解决了传统的模糊C均值聚类算法对初始值敏感的问题,另一方面克服了传统的聚类算法对数据集进行等划分的缺陷。实验结果表明,该方法不仅具有良好的收敛性,而且还可以有效地把目标从背景中分割出来,具有重要的实际应用价值。

关键词: 模糊C均值聚类, 二维熵, 图像分割