摘要
均值滤波或中值滤波只对某一种噪声有较好的滤波效果,为了克服单一滤波的缺陷与不足,提出一种分类滤波的去噪新方法。该方法先判断像素点是否需要滤波处理;再对需要处理的像素点按噪声特点进行分类,然后采用一种基于模糊隶属度的加权裁剪均值滤波方法滤除平坦噪声,采用中值滤波滤除孤立噪声。实验证明此方法可以抑制对所有像素点的同一滤波处理而造成的图像边缘细节模糊,而且对滤除多种不同类型的噪声有较好效果。
The mean filtering or median filtering just has good filter effect on removing only one kind of noise. In order to overcome the deficiency of single filtering, this paper puts forward a new method of sorted filtering for removing noise. Before filtering, pixels should be decided whether they need filtering or not. Then pixels needing filtering are sorted by their different noise characteristics. Weighted trimmed mean filtering of fuzzy membership function is used to remove flat noise, and median filtering is to remove isolated noise. The experiment proved that this method can reduce image edge fuzziness caused by processing all pixels in a same way, and it makes good effect on filtering many different kinds of noises.
出处
《计算机仿真》
CSCD
2007年第9期187-190,共4页
Computer Simulation
关键词
分类
均值滤波
中值滤波
模糊隶属度
加权
Sort
Mean filtering
Median filtering
Fuzzy membership
Weighted
