邻域均值检测的迭代加权中值滤波算法  

Iterative weighted median filter based on detection with mean of neighboring pixels

作　　者：陈家益 战荫伟[2] 曹会英 董梦艺 Chen Jiayi;Zhan Yinwei;Cao Huiying;Dong Mengyi(School of Biomedical Engineering,Guangdong Medical University,Zhanjiang Guangdong 524023,China;School of Computer Science&Technology,Guangdong University of Technology,Guangzhou 510006,China;Second Clinical Medical College,Southern Medical University,Guangzhou 510515,China)

机构地区：[1]广东医科大学生物医学工程学院,广东湛江524023 [2]广东工业大学计算机学院,广州510006 [3]南方医科大学第二临床学院,广州510515

出　　处：《计算机应用研究》2020年第6期1906-1909,1915,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(61170320);广东省科技计划资助项目(2017B010110015);广州市科技计划资助项目(201604016034);湛江市科技攻关计划项目(2017B01142)。

摘　　要：针对现有滤波算法在噪声检测与去除上存在的相应缺陷,提出了邻域均值检测的迭代加权中值滤波算法,对噪声检测与去除方法分别进行改进。算法根据噪声的灰度特征进行噪声检测,再基于邻域像素的相关性,用邻域的均值作进一步的检测;运用基于高斯曲面的加权算子,以迭代的方式,用邻域中信号像素的加权中值对噪声进行去除。实验结果证明,相对于现有滤波算法,所提算法具有更好的去噪性能,在保持高信噪比的同时,能很好地保持图像的纹理结构。The existing filters have defects in noise detection and removal. In view of these problems,this paper proposed an iterative weighted median filter based on detection with the mean of neighboring pixels,which aimed at improving the techniques of noise detection and removal. This proposed filter performed noise detection by the intensity characteristic of impulse noise,and then,taking full advantage of the correlation of neighboring pixels,performed further noise detection by the mean value of neighboring pixels. By employing the weighted operator which was derived by Gaussian surface,it used the weighted median of neighboring noise free pixels for noise removal iteratively. The experimental results show that the proposed filter outperforms the state-of-the-art filters by achieving belter denoising performance,and preserves well the texture structures of image while holding a high signal-to-noise ratio.

关 键 词：图像去噪 噪声检测 中值滤波 邻域均值检测 迭代加权中值滤波 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]