基于二次检测和极值修剪的自适应滤波算法  

Adaptive filtering algorithm based on two detection and extremum trimmed

作　　者：陈家益 熊刚强 战荫伟[2] 曹会英 CHEN Jiayi;XIONG Gangqiang;ZHAN Yinwei;CAO Huiying(School of Information Engineering,Guangdong Medical University,Zhanjiang 524023,China;School of Computer Science and Technology,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]广东医科大学信息工程学院,广东湛江524023 [2]广东工业大学计算机学院,广东广州510006

出　　处：《安徽大学学报（自然科学版）》2019年第5期35-40,共6页Journal of Anhui University(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61170320,11347150);广东省自然科学基金资助项目(S2011040002981,2015A030310178,2014A030310239)

摘　　要：当图像的噪声密度提高时,现行的中值滤波算法的滤波性能递减,滤波图像出现严重的失真.针对现行的中值滤波算法的局限性,基于二次检测和极值修剪的自适应滤波算法,对噪声检测和噪声滤除两方面分别进行改进.对于噪声检测,用算法预先设置的最大滤波窗口的灰度极值进行噪声检测,对检测出来的可疑噪声,用修剪灰度极值后的最大滤波窗口的中值做进一步的噪声检测.对于噪声点,自适应地用修剪灰度极值后的滤波窗口的中值取代.滤波实验主客观两方面的结果充分证明:相对于现行的中值滤波算法,论文算法有着更加良好的滤波效果,在滤除噪声的同时,很好地保持了图像的边缘和细节部分.The filtering performance of the current median filtering algorithms decreased and the filtered image was seriously distorted,when the noise density of image increased.Against the limitations of the current median filtering algorithms,the adaptive filtering algorithm based on two detections and extremum trimmed improved the noise detection and noise filtering respectively.This new filtering algorithm detected the noise with the gray extremum of maximum filtering window which was set in advance,then carried out further noise detection on the suspicious noise discovered by the gray extremum with the median of maximum filtering window after trimming the gray extremum.The filtering algorithm adaptively replaced the noise with the median of filtering window after trimming the gray extremum.Compared with the current median filtering algorithms,the result of both subjective and objective of the filtering experiment demonstrated that this algorithm had a significant better filtering performance,preserved the edges and details of image in a good shape while filtering the noise.

关 键 词：图像滤波 中值滤波 二次噪声检测 极值修剪 自适应滤波算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]