北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2005, Vol. 28 ›› Issue (3): 59-61.doi: 10.13190/jbupt.200503.59.wangjy

• 论文 • 上一篇    下一篇

基于2级检测的脉冲噪声滤除算法

王建勇,周晓光,廖启征   

  1. 北京邮电大学 自动化学院, 北京 100876
  • 出版日期:2005-06-28 发布日期:2005-06-28

Algorithm for the Removal of Impulse Noise Based on Two Stage Detection

WANG Jianyong,ZHOU Xiaoguang,LIAO Qizheng   

  1. School of Automation, Beijing University of Posts and Telecommunications, Beijing 100876, China
  • Online:2005-06-28 Published:2005-06-28

摘要:

以图像结构和噪声的特征分析为基础,提出了一种新的脉冲噪声滤除算法. 算法在噪声检测阶段采用2级开关检测,即先以极值法进行初步检测,然后利用像素邻域的结构信息,并借助设定的阈值识别出该点的性质(信号点或噪声点). 最后以滤波窗口内像素的灰度排序均值(ROM)替换噪声点的灰度值. 实验表明,本算法在滤除固定值脉冲噪声的同时可以有效地保护图像细节.

关键词: 图像处理, 脉冲噪声, 中值滤波, 噪声检测, 噪声率

Abstract:

A new algorithm of impulse noise removal is proposed based on the characters of image structure and noise. During noise detection a switching method which have two stages is adopted. The min-max operator is used first. Then further detection is conducted by using the information of structure around the pixel. The pixel is identified between noise and signal by a threshold. If it is a noise point, the value is replaced by rank ordered mean (ROM) of the filter window during noise removal. Compared with other several algorithms, the proposed algorithm can protect details of image efficiently while removing fixed valued impulse noise.

Key words: image processing, impulse noise, median filter, noise detection, noise ratio

中图分类号: