Calculating color sensations from arrays of physical stimuli

Complex real-life images recorded in tungsten, daylight, and skylight illuminants were digitized and displayed on a computer-driven color television monitor. The displayed images were calibrated so as to reproduce the gray scale and color balance of the original scenes. Observers made matches to a series of color squares in these images. The matching stimulus was an adjustable colored square in a constant white surround. Using algorithms based on Retinex color theory and the reset ratio-product model for lightness, the digitized images were processed to obtain predictions of color sensation for the test squares. The calculation did not require a priori information about the nature of the illumination. Predicted color sensations were found to be in reasonable agreement with the observers' matches.