A median filter of image based on grey clustering

According to the deficiencies of traditional median filter algorithms and the characteristics of salt & pepper noise, this paper puts forward an algorithm of the clustering median filter. The gray value of the image block in all directions is computed statistically and analyzed to determine the relationship between the pixel and its neighborhood using cluster analysis in the algorithm proposed. According to the clustering and statistical analysis, whether a point belongs to the noise class or not is determined. If so, the point is processed with a particular median filter. The result of experiments shows that the proposed algorithm can filter out salt & pepper noise more effectively and can preserve image details even better compared the traditional algorithms.