一种自适应双阈值模糊中值滤波算法的研究  

An Adaptive Double Threshold Fuzzy Median Filtering Algorithm

作　　者：邵琪[1,2] 邬延辉[1,2] 薛培培[1,2] 

机构地区：[1]宁波大学信息科学与工程学院,浙江宁波315211 [2]宁波大学计算机应用技术研究所,浙江宁波315211

出　　处：《移动通信》2015年第8期75-79,共5页Mobile Communications

摘　　要：通过对图像处理中噪声过滤现状的研究,特别是针对椒盐噪声的处理,介绍了一种自适应双阈值模糊中值滤波算法。研究了该算法的整体流程,主要步骤是用滤波窗口处理图像时,计算出滤波窗口中像素的最小值、最大值、中值与平均值。把当前像素值与中值的差值作为模糊系统的输入,并设定两个阈值,小于最小阈值表明该点不是噪声点,介于两阈值间认为是轻度污染,然后利用隶属函数计算加权系数,代入去模函数去掉模糊。当大于最大阈值时,该点已严重污染,根据该点邻域已处理的像素点求均值。实验证实了该算法比其他去噪算法的效果好。Based on the research status of noise ifltering in image processing, especially salt and pepper noise processing, an adaptive double threshold fuzzy median ifltering algorithm was introduced and its overall lfow was analyzed. Its main processing steps are presented below. Image is processed by ifltering window to calculate the minimum, maximum, median and mean values of the pixels in the ifltering window. The difference between the current pixel value and the median value is used as the input of fuzzy system, and then two thresholds are set. If the difference is less than the minimum threshold, the point is not a noise point. If the difference is between two thresholds, it is slightly polluted. Membership function is used to calculate weighted coefficients, and then these coefficients are plugged to eliminate blur mode function. If the difference is greater than the maximum threshold, the point is seriously polluted and a mean value is calculated in the light of the processed pixels in its neighborhood. Experiments conifrmed the presented algorithm has better performance than other de-noising algorithms.

关 键 词：椒盐噪声 双阈值 隶属函数 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]