基于梯度的自适应椒盐噪声滤波算法

针对传统中值滤波方法不能有效保持图像细节信息和对图像适应能力差的问题,提出一种改进的椒盐噪声滤除算法。算法基于先检测、后滤波的思想,根据图像的极小梯度矩阵自适应计算噪声阈值,提高了噪声检测的准确性;为了更好地保持图像细节,对检测出的噪声像素进行多窗口中值滤波。多组去噪实验表明：所提算法对污染程度不同的图像具有良好的适应性,在滤除噪声的同时还可以有效还原图像边缘等细节信息。

Grad based self-adapt image filtering algorithm for salt and pepper noise

An improved image filtering algorithm on the basis of grad for salt and pepper noise was presented in this paper. The algorithm is developed by two steps. Firstly, the algorithm computed adaptively noise threshold based on the grey grad matrix of the input image. Then the advanced veracity of the noise was detected. To improve the image quality, the multi-window median value was introduced to filter noise pixels. The experimental results show that the proposed algorithm has a good performance in noise filtering,and also can restore the edges of the original image well.