中文摘要：

针对传统自适应中值滤波算法的不足，提出了一种改进的自适应中值滤波算法，该算法引入了角度和相关性，第一，根据角度α的大小，判断当前像素点是否为可疑噪声点，再由其像素间的相关性判断此可疑噪声点是否为噪声点。第二，把排序得到的中值与窗口中所有像素点的均值加权得到新的灰度中值，再将噪声点用新的灰度中值替代，从而达到滤除噪声的效果。实验结果表明，该算法滤波，既能有效地平滑噪声，又能保存细节，效果远优于传统自适应中值滤波算法。

英文摘要：

Aiming at the deficiency of traditional adaptive median filtering algorithm, a modified adaptive median filter algorithm is proposed. Angle and correlation are introduced in this algorithm. First, accord- ing to the size of Angle alpha, through the correlation between ond, new gray value is acquired to judge whether the current pixel is suspicious noise point, and again the pixels, to judge whether the suspicious noise is the noise point. Sec- through weighting the ranked median value and mean value of all pixels in the window, and then noise point is replaced by new gray value, thus to achieve the effect of noise remov- al. Experimental results show that the algorithm could effectively smooth noise while preserving the details, and the effect is far superior to that of the traditional adaptive median filter algorithm.