Studies of practical daylight simulators for industrial colour quality control

Daylight simulators are widely used in industries for visual assessments and internal light sources of colour measuring instruments. As the CIE defined its standard daylight illuminants based on a combination of various measurement results, the precise realisation of CIE daylight illuminants is virtually impossible. Daylight simulators have therefore been developed to simulate the CIE daylight illuminants only approximately. The variety of daylight simulators used in practice caused a big concern on the quality of daylight simulators for industrial colour quality control. This study aims to investigate the variations of practical daylight simulators and the performances of standard methods including the BS950 band value method, CIE metamerism index method and CIE colour rendering index method for evaluating daylight simulators. The study also aims to reveal the discrepancies between various standards for specifying daylight simulators and to highlight the influence of these discrepancies on industrial colour quality control. An industrial survey on viewing cabinets was first carried out. The variations of D65 and D50 simulators accumulated were analysed in terms of colorimetric and spectral results. The results show that the D65 simulators generally performed better than the D50 simulators with higher CIE metamerism index and colour rendering ratings. The filtered tungsten lamps exhibit the best quality while the three-band fluorescent lamps show the worst quality. It was also found that both the CIE and ISO might not give appropriate specifications for the chromaticities of daylight simulators, i. e. too lenient for the former and too strict for the latter. A psychophysical experiment using real metameric pairs was conducted for evaluating the quality of six D65 daylight simulators. It was found that in general, all the simulators studied agreed well with each other in terms of the visual results, except for the three-band fluorescent lamp. The results show that BS 950 band value method and CIE metamerism index method are equally reliable for evaluating the quality of daylight simulators. It was also revealed that a simulator having band-value deviations well below the BS 950 tolerance corresponds to a high CIE metamerism index rating and therefore is judged as good quality. Seven metamer sets were generated based on different principles. These metamer sets were used to compare various metamerism indices for quantifying metamerism as well as to evaluate two methods, goodness-of-fit of SPD and CIE metamerism index for assessing daylight simulators. The results show that the measure of goodness-of-fit of SPD does not agree well with the CIE metamerism index method. It was proved that the CIE metamers are representatives of real metamers, however, the CIE metamer set gives smaller colour differences for the test simulators because it exhibits lower metamerism degree comparing to the generated metamer sets. It was also found that a limited number of metamers could be selected from a range of real metamers to perform as effective as the CIE set for evaluating daylight simulators.
The CIE colour rendering index method for evaluating daylight simulators was investigated using the CIE test colours and new sets of test colours. It was found that the CIE test colour sets agreed better with the paint sample sets than with the textile or thread sample sets. The four colour difference formulae, CIELAB, CMC, CIE94 and CIEDE2000, exhibit a similar performance for 'calculating CIE colour rendering index, and they all outperform CIEU*V*W*. No significant difference was found for the performance between two chromatic adaptation transforms, von Kries and CMCCAT2000. The results also show that it might be more appropriate to adopt the total CIE test colours instead of the first eight colours for calculating CIE colour rendering index. A new set of test colours, selected from the CIE' and other test colours for showing medium to large colour inconstancy as well as covering a large colour gamut, was proved to have a better performance than the CIE test colours. Finally, a method was developed for optimising the spectral power distributions of daylight simulators. Significant improvements' in quality were' achieved for the test lamps after the optimisation. A guideline for viewing cabinet design was also proposed based on the relevant standards and information accumulated from the industrial survey on viewing cabinets.