计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (36): 208-210.

• 图形、图像、模式识别 • 上一篇    下一篇

考虑均值的滤除脉冲噪声的自适应非线性滤波

贾惠珍1,王同罕2   

  1. 1.广西师范大学 计算机科学与信息工程学院,广西 桂林 541004
    2.江苏大学 计算机科学与通信工程学院,江苏 镇江 212013
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-12-21 发布日期:2011-12-21

Adaptive nonlinear filter for removing impulse noise based on mean

JIA Huizhen1,WANG Tonghan2   

  1. 1.School of Computer Science and Information Technology,Guangxi Normal University,Guilin,Guangxi 541004,China
    2.School of Computer Science and Telecommunication Engineering,Jiangsu University,Zhenjiang,Jiangsu 212013,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-12-21 Published:2011-12-21

摘要: 针对脉冲噪声滤除,提出一种基于均值的自适应非线性滤波。其算法思想是先进行噪声检测并生成相应的噪声标志矩阵,随后扫描噪声标志矩阵,对信号点则直接输出,而噪声点则根据窗口内噪声点个数自适应选择滤波窗口,接着生成一个由滤波窗口各像素灰度与相应均值之差形成的差别矩阵,并赋予差别矩阵各元素不同的权重,最后返回加权后差别矩阵中最小值的位置,并用该像素的灰度取代噪声点。噪声点滤除算法,其实质是找出一个空间位置上离噪声点最近且最接近均值的像素的灰度来取代噪声点。通过实验,该方法具有更好的综合滤波性能。

关键词: 脉冲噪声, 噪声标志矩阵, 自适应, 均值, 差别阵, 加权

Abstract: An adaptive nonlinear filter for removing impulse noise based on mean is proposed in the paper.The method works in this way:by noise detection,the pixels in image are divided into noise points and signal points and a noise label matrix is generated.It scans the noise label matrix,if the pixel is a signal point,the pixel intensity is directly input,otherwise,the size of filtering window is adaptively adjusted to the number of noise points in the window.It generates a difference matrix which is different between the pixel intensity in the filtering window and the mean of the pixel intensity in the filtering window,adaptively assigns weight value to the pixels of the difference matrix,and returns the min pixel of the weighted difference matrix.Its intensity replaces the intensity of the noise point.The algorithm of removing noise point finds the pixel which the position is near to the noise point in the space and the intensity is nearly to the mean of the pixels in the filtering window,then the pixel intensity replaces the noise intensity.The experiment shows that the filter has better filtering performance.

Key words: impulse noise, noise label matrix, adaptive, mean, difference matrix, weight