Using recovered reflectance to predict color

Predicting a surface color under different lights is an easy task if the surface reflectance function is available. However, often only colorimetric information is available, and the tristimulus values of a color that undergoes an illuminant change are estimated using transforms inspired by the von Kries coefficient rule model. We propose a new method, based on the reconstruction of multispectral data, for modeling illuminant change. It assumes that the problem is specific for a domain of colors, and that this domain can be modeled in a three dimensional Gaussian space. The performance of our method is compared with that of the simple von Kries diagonal transform, and the results are reported for three sets of data.