中文摘要：

针对传统自适应中值滤波在保存图片边缘细节方面的不足，提出了一种改进自适应中值滤波算法，在有效去除图像中高密度脉冲噪声的同时保证图像的边缘细节信息。首先利用对可疑噪点进行梯度方向连续求导以区分该点是噪声点或边缘点。然后，对噪声点还原值的求取同时考虑到中值与均值。中值与均值的计算为避免被强噪声干扰，先将滤波窗口中的重复值精简为一个，后在进行中值和均值的运算。最后通过仿真实验，结果表明该算法在保存图像边缘细节方面显著优于传统自适应中值滤波算法。

英文摘要：

Aiming at the defects of traditional adaptive median filter in preserving the details of image edge, a modified adaptive median filtering algorithm is proposed, which could effectively remove the high density impulse noise in the image and ensure the edge details of the image. Firstly, the derivative of that point a- long the gradient direction is acquired so as to distinguish the point from the noise or the edge. Then, the value of the noise point reduction is taken into account in the mean value and the median value. In order to prevent them from being disturbed by strong noises, the number of repetition value in the window is re- duced to one, and then the mean value and median value operation is done. Finally, the simulation exper- iments show that this algorithm is clearly superior to traditional adaptive median filtering algorithm in pre- serving the image edge details.