Illuminant color estimation method for the scene illuminated by several illuminants is proposed. In the past, most of the conventional methods focused on the scene under one illuminant and assumed the color of the illuminant is constant throughout the scene, however, these are not always the case. These days, the methods for estimating several colors of the scene illuminants are studied. Some of them are based on dichromatic reflection model, and assume that colors of the illuminant in the small region in the image is only one and constant throughout the region. Proposed method is based on Gray-World assumption, which doesn't need the specular reflection components but variety of colors in the image. In the method, image is divided into small regions and the illuminant colors are estimated for the small regions after the judgement whether each small region in the image includes variety of colors is or not. By clustering the estimation for each small region, final result is derived. In the experiments, simulated images under two illuminants are used and the estimated results are evaluated. The results show that the estimation error by the proposed method is smaller than those by the conventional one.
Harumi Kawamura, "Illuminant Color Estimation in an Image under Several Illuminants Based on Gray-World Assumption" in Proc. IS&T Int’l. Symp. on Electronic Imaging: Color Imaging XXIII: Displaying, Processing, Hardcopy, and Applications, 2018, pp 295-1 - 295-4, https://doi.org/10.2352/ISSN.2470-1173.2018.16.COLOR-295