A re-coloring algorithm for a color image using statistic scheme in CIE L⋆a⋆b⋆ color space

This paper presents a simple and efficiency algorithm for re-colorizing a color image. This algorithm transforms the distribution of the Lab value of the original image and the distribution of the Lab value of the reference image into Gaussian distributions, respectively. In each plane (L, a, and b), the multiple regression is used to find the best fitting functions for the transformed data. The re-colorized transformation data are then calculated by the best fitting functions. Finally, the re-colorized data is obtained by taking the inverse transformation of the re-colorized transformation data. The algorithm is tested under various input images. The experimental results show that the presented algorithm can effectively re-colorize a color image into another color hue (e.g. from cool hue into a worm hue).