Image smoothing based on the mean-shift algorithm

A mean-shift algorithm based on the NTSC standard YIQ color space is proposed to reduce noise and better preserve the edges of images.In this method,the image is first transformed from RGB space to YIQ space.Because the correlation of the three components in YIQ space is weak,Y,I,and Q can be separately smoothed with the mean-shift algorithm according to their different step sizes.Then the smoothed image is transformed from YIQ space to RGB space,thus resolving the problem that the R,G,and B cannot be independently smoothed owing to strong correlations between R,G,and B.The convergence of mean shifting points was strictly proven.Experimental results prove that this algorithm can better meet smoothing requirements than Wiener filtering or the mean-shift method based on RGB space.