Techniques for simulating object color under changing illuminant conditions on electronic displays

Abstract

Abstract We compare several computational schemes that can be used for the simulation of object colors under changing illuminant conditions on electronic displays. These schemes differ in the assumptions that are made concerning illuminant invariant quantities, and the amount of a priori information that is required. This latter feature is especially important for computer graphics applications and theoretical color constancy research because in many instances information about illuminant and object characteristics is minimal. A large set of tristimulus values of Munsell papers, as defined under CIE Standard Illuminant C, served as input data. Next, illuminant invariant quantities were obtained for each computational scheme that were used to predict x, y chromaticity coordinates and relative luminosity Y under nine different phases of daylight, varying from 4000 K to 20000 K. It was found that the differences between so‐called reflectance models on the one hand, and theoretically less accurate tristimulus ratio models on the other, are small and in most cases within three MacAdam units of color difference ≈1 JND).