A Modified Development of CIE Color Difference Formula via Polynomial Regression Approaches

A modification of color difference formula based on CIEDE2000 is developed. A residual weighting factor was acquired to enhance the predictability of CIE color-difference formula effectively. The residuals, ¿E', were calculated between CIEDE2000 difference values, ¿E
₀₀
, and visual results, ¿V. The weighting function was derived by using three types of polynomial regression, including RE-common, RE-divisional and RE-multiple. The coefficients of polynomial equation were determined to predict the minimized residuals, ¿E
_P
'. The modified color difference values, ¿E
_RE
, were obtained by calculating ¿E
₀₀
with ¿E
_P
'. A combined experimented dataset (COM) was formed, which was consisted of four reliable color-difference datasets. The equation was tested together with the other CIE color-difference formulae using the combined dataset and each of blue, gray and dark datasets. The results showed that the modified formula performed better than CIEDE2000.