一种有效的去除图像混合噪声的滤波算法

A Valid Filtering Algorithm for Removing Mixed Noise in Image

依据噪声特点以及图像的像素关联性,提出了一种有效的数字图像混合噪声滤波算法。该算法首先对图像像素进行检测,依据受脉冲噪声污染的图像像素与其周围邻域多数像素在亮度上具有显著差异的特点准确检测出脉冲噪声,然后以检测到的脉冲噪声为被处理像素,用其邻域内未被脉冲噪声污染的图像像素对其进行中值滤波,由于当前脉冲噪声点以及邻域脉冲噪声点均未参与滤波运算,从而较好地保护了图像的细节。对图像中的高斯噪声,提出了一种改进的加权均值滤波算法,该算法在定义相关度函数时,既考虑了像素的灰度相关性,又考虑了像素的位置相关性。实验结果表明:提出的混合噪声滤波算法不仅可以有效滤除图像中的混合噪声,而且还可以较好地保护图像的细节。

According to the characteristics of noise and the pixel correlation of image, a valid filtering algorithm for removing mixed noise in digital image is proposed in this paper. The algorithm can detect impulse noise accurately according to the characteristics that image pixels contaminated by impulse noise are significantly different from most of the pixels of its neighborhood in brightness, and then the authors tooe the median filtering algorithm to filter the detected impulse noise using those pixels which are not contaminated in filtering window. Because the processed impulse noise and it＇s neighborhood impulse noise are all not attended filtering operation, thus a good protection of the detail of the image was implemented. Afterwards, for removing Gaussian noise in image, the paper proposed an improved weighted average filtering algorithm. The algorithm considers not only gray correlation but also position correlation of pixels in defining correlation function. The experimental results show that the proposed mixed filtering algorithm not only can filter out the mixed noise in image effectively but also can protect image detail well.