Robust Estimation of Skin Pigmentation From Facial Color Images Based On Color Constancy

pigment separation algorithms based on independent component analysis (ICA) have been proposed. Their key limitation lies in not only the lack of robust estimation of skin pigmentation, melanin and hemoglobin, under different illumination, but also the inefficiency of the algorithm. In this paper, we propose a new method for estimating the concentration of skin pigmentation, combining with the known color constancy algorithm which is applied to the skin image as the preprocessing to remove the color interference from illumination. And as an improvement of the conventional pigment separation algorithm, we propose a new verification rule of pure color intensities matrix, discarding the complex computations of distinguishing the vectors of the pure color intensities matrix which correspond to melanin and hemoglobin components. Finally, we qualitatively evaluate the concentration distribution corresponding to melanin and hemoglobin, and compare the results of before and after using the color constancy algorithm. The experiments show that our algorithm obtains comparable results as the state-of-the-art pigment separation methods with the merit of higher efficiency and robustness against change in illumination.