一种新的自适应中值滤波方法。将3 x3窗口中心的极值点作为候选噪声点，若候选噪声点仍然是7 x7窗口的极值点，则该点即是噪声点。若以噪声点为中心的3 x3滤波窗口的中值不是噪声，则噪声用中值替换。重复以上过程，直到没有噪声点被替换。如果图像中仍然存在人的噪声团块，则噪声用相邻的三个信号点的灰度均值替换。实验结果表明，该方法能够有效去除脉冲噪声，并在抑制噪声的同时很好地保护图像的细节。-A new adaptive median filter. The extreme points of the 3 x3 window center point as a candidate noise, if the candidate noise point is still the extreme points of the 7 x7 window, the noise points. If the noise point of the 3 x3 filter window median is not noise, the noise replaced with the median. Repeat the above process until there is no noise points to be replaced. If the image still exists in clumps of the noise, the noise replace the gray average of the three adjacent signal points. The experimental results show that this method can effectively remove the impulse noise, and noise suppression to protect the details of the image.