Study Of Color Vision Discrimination Characteristics And Color-difference Evaluation

The quality of color is being of greater importance with the development of science and technology and the improvement of the living standard.In order for scientific and effective description and evaluation of colors and their differences,the color vision discrimination characteristics should be comprehended.Though several uniform color spaces and color-difference formulae have been proposed based on the related visual experimental datasets,significant differences are found between these datasets,so the color discrimination characteristics and the corresponding color prediction models must be systematically tested and analyzed.In addition,parametric effect upon the color vision discrimination characteristics needs to be further discussed.Hereby,using CRT-generated color stimuli by the psychophysical visual experimental methods in this study,the color discrimination threshold and suprathreshold color-difference data were obtained to analyze the color vision discrimination characteristics of human eyes, which were used to compare and evaluate the prediction performances of the CIELAB-based typical color-difference foumulae,namely,CIELAB,CMC,BFD, CIE94,LCD and CIEDE2000,then the lightness-,chroma- and hue-weighting functions of the color-difference formulae were optimized,respectively,to improve their color-difference prediction performances.The color discrimination thresholds of nine CIE color centers on the a~*b~*,a~*L~* and b~*L~* planes in the CIELAB color space were measured using the interleaved staircase method,which provided a visual dataset at the threshold level based on self-luminance colors for the investigation of color vision discrimination characteristics.The data analysis indicates that CIELAB color space is not uniform for color discrimination thresholds.The fitted chromaticity ellipses were employed to estimate the prediction performances of the color-difference formulae,in their original and optimized forms,respectively,which shows that the CIEDE2000 formula performs the best in the both forms in the whole color space.The lightness-,chroma-, and hue-weighting functions were also evaluated,which reveals that CMC has the best lightness prediction performance,while LCD outperforms other formulae in predicting the chroma and hue differences.Based on the principle of the constant stimuli method,the suprathrehsold color-difference visual data at the 19 color centers on the a~*b~*,a~*L~* and b~*L~* planes in the CIELAB space were obtained,which produced an accurate color-difference dataset covering the whole color gamut of the CRT display.Detailed analysis of the corresponding color vision characteristics implies that the CIELAB space is also not visually uniform for suprathreshold color differences.The original and optimized color-difference formulae and their weighting functions were compared,respectively, which indicates that,as a whole,CIEDE2000 has the best prediction performance, followed by LCD.For the lightness-difference prediction CIELAB and CIE94 perform the best,and LCD outperforms others for the chroma- and hue-difference prediction.The impact of different visual scales and different perceptual correlates on the color vision discrimination characteristics of the visual tolerances were investigated, and there are few similar researches presently in the world.The lightness,chroma and hue tolerances were acquired based on different visual scales and different perceptual correlates using the psychophysical method of constant stimuli,which were applied to the analysis of the color vision discrimination characteristics and the evaluation of the prediction performance of the color-difference weighting functions.For the visual tolerances with different perceptual correlates,there is a considerable crispening effect between lightness tolerances and lightness scales,while lightness tolerances have no obvious correlation with chroma or hue angle.Chroma and hue tolerances show some hue-angle dependencies,and inverse and direct proportions to lightness and chroma values as well,respectively.For visual tolerances with different visual scales,the relations between visual tolerances and visual scales are nearly linear. Among the tested typical color-difference formulae,CIEDE2000 outperforms others for the lightness- and hue-difference prediction,while LCD has the best chroma-difference prediction performance.Based on the visual datasets of CRT colors and the combined datasets of CRT colors and surface colors,respectively,the lightness-,chroma-,and hue-weighting functions of the color-difference formula were optimized to propose the corresponding weighting equations with the consideration of the correlations between visual tolerances and perceptual correlates.The prediction performance testing results of these optimized weighting functions indicates that there are considerable discrepancies between the visual data of CRT colors and surface colors,which implies that different weighting functions may be adopted to evaluate the color differences for different color media.