Bayesian approach for developing threshold color-difference models by the strip-pair comparison method

Abstract

[EN] A Bayesian approach alternative to the one used in the strip-pair comparison method for developing threshold color-difference models is presented in this paper. Strip-pair comparison method is based on the construction of color-control strips made of pairs of patches put in contact and ordered by increasing the CIELAB color difference. Observers are required to indicate the number of the pair of patches in every strip for which they begin to perceive a just noticeable color difference. Frequency data obtained, from repeating several times the visual assessment, is recorded to build a Bayesian multinomial logistic regression model, which allows the determination of the coefficients of the color discrimination ellipsoids. The results of the Bayesian approach agree closely with the results obtained to validate strip-pair comparison method for the same theoretical frequency data. The main advantage of the Bayesian approach over many other methods is that it allows a direct analysis of the statistical variability of the estimated parameters by means of confidence intervals and other measures of statistical variability.