Abstract

We developed a model using the critical axes from Boynton and Kambe to describe the discrepancy in purity-discrimination data measured from white, ΔP(white), and from the spectrum, ΔP(spectrum). In this study, we used a computer-controlled four-channel Maxwellian view system to investigate the model predictions. For five normal observers, we measured chromatic discrimination steps along the tritan and R/G axis, purity-discrimination functions from white (0.3536,0.3545), and purity-discrimination functions from the spectrum. A double-random staircase, temporal forced-choice paradigm was used. The luminance levels varied from 2.86 to 286 td. In confirmation of the model, the ΔP(white) thresholds were similar to classical data, and the ΔP(spectrum) thresholds at a fixed luminance level depended on the individual observer's tritan discrimination threshold. The data also showed that test luminance level had a major effect on ΔP(spectrum) but only a minor effect on ΔP (white). Furthermore, we observed that a common TVI template cannot describe discrimination changes with both chromaticity and mean luminance. We refined the chromatic discrimination model to account for the main trends of the colorimetric purity data.

© 1990 Optical Society of America