基于改进中值滤波的图像噪声去除算法的研究

Study on the image denoising based on the improved median filtering algorithm

为了有效且快速滤除高密度椒盐噪声,同时又能很好地保护图像的细节,提出了一种基于中值滤波和均值滤波的改进滤波算法。该算法首先通过噪声检测过程确定图像中的噪声点和非噪声点,然后利用小窗口进行迭代滤波,根据窗口内非噪声点及噪声点数目的不同,采用相应的算法对窗口内噪声点进行适时的更新。试验结果表明,该算法能有效滤除密度从5%至80%的分布范围内的椒盐噪声,在图像的主观质量和客观质量方面均取得了较好的效果,有利于图像进一步的分析、判读。

In order to remove the high-density salt and pepper noise,and preserve the details of the original image well,an improved filtering method was proposed based on the median filter and the mean filter in this paper. In the first phase,it was necessary to identify pixels which were likely to be noise. In the second phase,the image was filtered recursively by the small window,salt and pepper pixel in the current window was calculated,then the noise pixel using the median and mean weighted value of the non-noise pixel were updated. The experimental results showed that the proposed method behaved excellently for all noise ratios,from 5% to 80%. This method has achieved the purpose of the good filtering effect on the subjective quality and objective quality,which was beneficial for the further analysis and interpretation of images.