Color Calibration Method for Images Taken by Different Imaging Conditions Which Suppresses False Color

Owing to different imaging conditions, color reproduction of images is usually different; hence, color calibration methods have been developed to unify color reproduction. Iterative distribution transfer (IDT) is one such method that can achieve good color calibration for most pixels of an input image by matching the color distribution of a reference image. This method, however, may lead to unnatural colors, called false colors. In this study, we propose a new color calibration method, which effectively suppresses false colors, whose result is similar to that obtained via the IDT. The input image is projected to specific powers called bases, which are multiplied by the corresponding projection coefficients and added to obtain the projection result. The projection coefficients can be obtained by minimizing the objective function, which is the mean-square error between the projection result and the IDT result. Considering that a false color in the IDT results is expected to be similar to the color of an adjacent pixel, a color-similarity-related item is added to the objective function to address this. Color similarity can be measured by the chromatic aberration between the adjacent pixels of the input image. By ensuring that similar colors in the input image remain similar in the output image, false colors can be suppressed. For the experiment, we use false color index, Kullback-Leibler divergence, and visual evaluation to evaluate the calibration results, which consistently proves that this method offers a better color calibration effect compared with other methods.