Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (16): 163-167.

• 图形、图像、模式识别 • Previous Articles     Next Articles

Method of image region segmentation based on K-means clustering algorithm

LI Su-mei1,2,HAN Guo-qiang2   

  1. 1.Educational Technology Center,Guangdong University of Foreign Studies,Guangzhou 510420,China
    2.College of Information Science & Engineering,Central South University,Guangzhou 510641,China
  • Received:2008-01-11 Revised:2008-04-03 Online:2008-06-01 Published:2008-06-01
  • Contact: LI Su-mei

基于K-均值聚类算法的图像区域分割方法

李苏梅1,2,韩国强2   

  1. 1.广东外语外贸大学 教育技术中心 广州 510420
    2.华南理工大学 计算机科学与工程学院 广州 510641
  • 通讯作者: 李苏梅

Abstract: A K-means alogorithm of oneself decideing the clustering number is proposed,and based on this alogorithm,a method of region-based image segmentation is introduced.For this method,a suitable color space is selected,the features of color,texture,and location are extracted,and the feature space is formed.Then,in this feature space,an image is clustering and separate into regions by proposed method.Finally,the features of regions are extracted.In this paper,the proposed method is detailedly introduced,and the experiment results and the comparision results with the similar approach are provided.Experiment results show the proposed method has the quickly segmentation speed and good sementation results,and it is fit for region-based image retrieval system and has the better applied values.

Key words: self-adaptive K-means clustering, region-based image segmentation, features of image regions

摘要: 提出了一种自动确定聚类数目的K-均值聚类算法,并基于这种算法介绍了一种彩色图像区域分割方法。这种方法首先选择合适的彩色空间,抽取图像的像素点颜色、纹理及位置等特征,形成特征向量空间;然后,在此特征空间中,运用提出的方法进行聚类和图像区域分割;最后,抽取图像区域的特征。对提出的方法进行了详细的介绍,给出实验结果分析,并与相类似的方法进行了比较实验。实验结果表明,提出的图像区域分割方法具有分割速度快、效果好等特点,适合于基于图像区域检索系统,具有较强的实用价值。

关键词: 自适应K-均值聚类, 图像区域分割, 图像区域特征