图像脉冲噪声滤波算法

Image impulse noise filtering algorithm

针对低噪声污染图像提出了一种改进的中值滤波算法。该算法通过计算滑动窗口内的像素均值和方差,根据数理统计特性,自适应选定阈值,对符合噪声条件的像素进行初次滤除,然后采用开关中值滤波算法对不符合条件的像素再次滤波。实验结果表明,该算法既能有效地去除噪声,又能清晰地保持图像边缘,降低了传统改进型中值滤波算法对阈值的依赖性和对图像边缘细节的损害程度,且滤波性能优于一些典型改进型中值滤波算法。

An improved median filtering method is proposed to suppress the impulse noise with low contamination ratios.It determines the threshold adaptively by calculating the average gray and variance of the pixels in the slide window according to the mathematical statistics characteristic,and replaces the pixels which are deemed as impulse noise by the median gray of slide window.Then uses the switch median filtering method to smooth the pixels which are not suit for noise condition in the former step again.Experiments indicate that proposed filtering method can remove the impulse noise effectively and preserve the image edge,furthermore,it brings down the dependency of the image threshold as well as the degree of damage of the image edge,and outperforms some typical improved median fil-tering method.