计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (11): 34-36.DOI: 10.3778/j.issn.1002-8331.2009.11.010

• 博士论坛 • 上一篇    下一篇

分类中值滤波算法的研究

黄 颖1,2,王卫星2   

  1. 1.重庆邮电大学 计算机科学与技术学院,重庆 400065
    2.电子科技大学 电子工程学院,成都 610054
  • 收稿日期:2008-12-11 修回日期:2009-02-13 出版日期:2009-04-11 发布日期:2009-04-11
  • 通讯作者: 黄 颖

Research on classification median filter algorithm

HUANG Ying1,2,WANG Wei-xing2   

  1. 1.College of Computer Science & Technology,Chongqing University of Posts & Telecommunications,Chongqing 400065,China
    2.School of Electronic Engineering,University of Electronic Science & Technology of China,Chengdu 610054,China
  • Received:2008-12-11 Revised:2009-02-13 Online:2009-04-11 Published:2009-04-11
  • Contact: HUANG Ying

摘要: 在标准中值滤波算法的基础上提出了基于聚类方法的滤波算法,该算法除了考虑当前像素和邻域组成的序列的统计特征之外,还考虑到了该序列的结构性特征。首先将该序列分成两类,然后计算当前像素所在的类别的中值作为滤波器的输出。该算法还设计了高低两个阈值来界定中值和均值之间的绝对差的范围,当绝对差属于不同的范围时采用不同的处理方法。最后使用了一些典型例子来说明该方法的去噪效果比标准的方法要更好一些。

关键词: 图像处理, 聚类方法, 中值滤波

Abstract: A modified median filter algorithm based on clustering method is presented in this paper.At first,the computation of median,mean of the series composed by the current pixel and its neighbors is necessary to estimate the distribution of the series,which is utilized to decide the classification method.Then,the series is classified into two classes with clustering method or fast classification method,and the median of the class which includes the current pixel is chosen as the result.Two parameters(T1 and T2) are used for the consideration of speed and accuracy.Some examples are given to prove the new algorithm can get better noise reduction effect.

Key words: image processing, clustering method, median filter