Abstract
Human color vision differs widely among individuals. This variability is problematic when trying to accurately reproduce colors across normal observers on different devices such as monitors or printers, especially with the advent of systems with more narrowband primaries. Thus there is increasing interest in measuring and correcting for the color profile of individual observers. Factors responsible for this variability are well known and include differences in lens and macular pigment density, LM cone ratios, photopigment peak sensitivities and optical density. Variability in these factors is reflected to various degrees in color matching and isoluminant settings. Color matching functions, though informative, are time-consuming, difficult to measure, and insensitive to cone ratios. In contrast, isoluminant settings are relatively easy to acquire, and are affected by LM cone spectra and ratios as well as preretinal screening, though largely insensitive to S cones. Here we model how well variations measured only in isoluminant settings – e.g. in a minimum motion task (Anstis and Cavanagh, 1983) - can recover the different sources of physiological variation and thereby predict an individuals’ color matches. Only two stimulus comparisons are required to define the isoluminant plane. We show that variations in lens and macular density (among the largest sources of variance in color matching), and LM ratios tilt the plane in distinct ways, and evaluate the extent to which these tilts can be used to estimate the values for these factors. We also evaluate the number of conditions (e.g. foveal vs. peripheral) and the stimulus spectra that are best for parceling out the different functions, and then compare how well these estimates can predict simulations of complete color matching functions. Our results help guide a determination of the smallest and most practical set of measurements that may suffice to characterize an individual’s color experience for reasonably accurate color reproduction.
Acknowledgement: supported by EY-10834 and P20-GM103650