混有高斯和脉冲噪声图像的一种新滤波方法

针对数字图像中的高斯噪声和脉冲噪声，提出了采用递推最小最大值法与中值滤波相结合的滤波方法。对数字图像首先采用递推最小最大值方法来除去脉冲噪声，然后采用中值滤波方法，并选择合适的窗口尺寸，可得到质量很好的图像。实验结果表明，递推最小最大值方法在抑制脉冲噪声的同时，还能保留良好的图像细节，特别是在多种脉冲噪声存在的条件下，效果更加明显，而中值滤波可以有效地除去剩余噪声，最终得到满意结果。

A New Filtering Method for Restoration of Digital Images Contaminated by Mixed Gaussian and Impulse Noise

A synthetic filtering method is proposed for removing mixed Gaussian and impulse noise, namely synthesizing recursive minimum-maximum method and median filter. At first, recursive minimum-maximum method is applied to remove impulse noise, then median filter to remove Gaussian noise and additional noise. As long as proper window size is selected, high-quality image can be restored. Experimental results indicate that applying recursive minimum-maximum method, fine details can be preserved and impulse noise can be suppressed effectively at the same time, especially under various impulse noise conditions, and the following median filter can remove the remaining noises effectively.