The influence of parametric effects on the appearance of small colour differences
The effects of the two physical parameters, background and gap, on the perception of small colour differences (AE* < 5) were investigated by use of 248 colour-difference pairs around 21 colour centres made from painted samples. Each pair was assessed by an average of 30 times under each viewing condition using a grey-scale method and/or a paired-comparison method. From the visual data, colour-difference ellipsoids (ellipses) or tolerances were determined by use of a logistic or a probit maximum-likelihood analysis model, or by a least-square method. The perceived tolerance sizes along the three colour-difference directions AL*, OC1 and AH` in the CIELAB space were found to be little influenced by a change of lightness of the grey background but significantly influenced (i. e., decrease in tolerance size) by a 0.51'g ap between a pair of samples. The gap factor for the lightness component was greater than that for chroma or hue components, both the latter having similar magnitudes. This could be an explanation for the increase of the lightness relative tolerance I (or parametric factor KL) by a factor of 2 in the three modified CIELAB formulae (CMC, BFD and CIE94) for acceptability judgements, in which textile samples, having an unclear dividing line between them, are mainly used. In addition, the value of the relative tolerance Q used is thought to be practically the ratio between the lightness and chroma tolerances. The experimental uncertainties from non-physical parameters were also quantified. The degrees of precision (i. e., standard error) of colour measurements and observer judgements were found to be good (±4% and ±7%, respectively). The different methods of scaling and data analysis were found to have little impact on the results. The lightness, chroma and hue tolerances with respect to the standard colour position in the CIELAB space were studied in detail using the various existing datasets and the set from this study. The lightness tolerance showed a clear dependency upon the metric lightness for medium to light colours, but in the case of dark colours there was a discrepancy between the datasets. Both the chroma and hue tolerances showed dependency upon both the chroma and hue-angle and not the single dependency upon the metric chroma, as assumed in the CIE94 formula. New weighting functions were derived from the above experimental evidence, and finally a new formula, LCD (Leeds Colour Difference) was proposed. The LCD formula is nearly as simple and flexible as CIE94 but smoothes the individual weighting functions compared to CMC and BFD, especially for lightness tolerances for light colours and chromaticity discrimination near the blue region. It was also found that the reliability of the BFD formula is improved when the size of the chroma weighting function is increased by 1.5 times and the form of lightness weighting function is made parallel to those of the other modified CIELAB formulae.