ZHOU Juan-Li, BANG Hai-Ou
The color migration effect of traditional methods depends on the subjective evaluation, and has a big application limitations. This paper proposes an improved method of image color transfer. It chooses the best matched color of the reference color area from the multiple source images by the brightness fitting, category selection. For the color images, the various regions of the target image are smoothed by using the probability of membership function, and the synthetic value of color transfer is computed by using the mean and variance of the reference color area; for the grayscale images, the components of reference and source color image are normalized, the gray level is divided, and the membership is weighted by the Sigmoid function, to achieve the color migration of gray images. Experimental results show that the improved method can generate convincing results with well performance in terms of objective evaluation or visual effects.