Color contrast modeling and CRTs

Abstract

Many researchers assume that the CIE uniform color spaces, CIELUV and CIELAB, can model color contrasts for CRT displays. These color spaces were developed for surface colors, not self-luminous colors as generated on CRTs. CIELUV and CIELAB combine, by vector summation, a luminance component (L*) and chromaticity components weighted by L*. The chromaticity components for CIELUV and CIELAB have limited areas of incompatibility when modeling data from self-luminous objects like MacAdam’s ellipses. The luminance component L*, a function of Y1/3, is based on the Munsell value function and has not been compared with data from self-luminous objects. We tested the accuracy of L* and Y1/3 in modeling the Weber fraction data. Over a typical CRT luminance range, Y1/3 is adequate while L* is not. Over an extended luminance range, neither Y1/3 or L* model the data. L* predicts the same number of just noticeable differences for any luminance range. A further problem is that previous psychophysical studies (e.g., MacAdam1) used vector summation of independent chromatic and luminance contrast components to model color contrasts, an approach not possible with CIELUV and CIELAB since L* appears in the chromaticity component.