Looking for the dichromatic version of a colour vision model

Abstract

Different hypotheses on the sensitivity of photoreceptors and post-receptoral mechanisms were introduced in different colour vision models to derive acceptable dichromatic versions. Models with one (Ingling and T'sou, Guth et al, Boynton) and two linear opponent stages (DeValois and DeValois) and with two non-linear opponent stages (ATD95) were used. The L- and M-cone sensitivities of red–green defectives were either set to zero (cone-loss hypothesis) or replaced by that of a different cone-type (cone-replacement hypothesis), whereas for tritanopes the S-cone sensitivity was always assumed to be zero. The opponent mechanisms were either left unchanged or nulled in one or in all the opponent stages. The dichromatic models obtained have been evaluated according to their performance in three tests: computation of the spectral sensitivity of the dichromatic perceptual mechanisms, prediction of the colour loci describing dichromatic appearance and prediction of the gamut of colours that dichromats perceive as normal subjects do.