Practical daylight sources for visual colorimetry

Abstract

This thesis was concerned with a systematic study of quantifying the quality of D65 simulator for practical visual colorimetry. The relationship between psychophysical data and quantification method for D65 simulator was studied theoretically and experimentally with an aim to establish a sound scientific understanding of the uncertainties, and hence to develop a mathematical model using colorimetric method to quantify the quality of artificial light sources with respect to visual assessment. An extensive literature review was carried out including the historical development of CIE standard daylight illuminant, the simulation of the CIE standard daylight illuminant, the typical standards in assessing the quality of daylight simulator and the applications of D65 simulator. The viewing conditions and psychophysical experiments were described. In addition, the previous studies in wavelength intersections of metameric pair and metamerism index were illustrated. The colorimetric calculations used in this study such as Lagrange formula, tristimulus value calculation methods and colour difference equations were discussed. From the literature review, it revealed that it is difficult in the simulation of CIE standard daylight illuminant. The use of various D65 simulators with different quality for visual evaluation resulted in the lack of precision in visual colour evaluation and might cause problems of illuminant metamerism. Further confusion in colour matching was introduced if metameric pairs are assessed using D65 simulators with a different spectral power distribution. However, there were some uncertainties in the usefulness of those quantification methods in assessing the quality of daylight simulator. In addition, little attention had been given in quantifying the quality of artificial light source for visual assessments directly. A survey of commercial colour viewing cabinets was carried out, which was focused on the types of D65 simulator and the numbers of artificial light sources used by them. The survey results showed that the number of artificial light sources used in the colour viewing cabinet was not the same. Moreover, the filtered halogen tungsten lamp and fluorescent lamp were the common technologies used to realize the CIE standard illuminant D65. In quantifying the quality of D65 simulator, fourteen D65 simulators used for visual evaluation were collected and their relative spectral power distributions were measured by tele-spectroradiometer. There were one filtered halogen tungsten lamp and thirteen fluorescent lamps. Two normalization methods used for D65 simulators were explored. The difference in the relative spectral power distributions between two types of D65 simulators was studied. Based on the results of spectroradiometeric measurements, the quality of each D65 simulator was evaluated by CIE No.51.2 method "A Method for Assessing the Quality of Daylight Simulator for Colorimetry" and 13.3 method "Method of Measuring and Specifying Colour Rendering Properties of Light Sources" respectively. The results showed that the quality of the D65 simulators were ranged from A to D categories in MI(vis) of CIE No.51.2 methods and the range of colour rendering index in CIE No. 13.3 methods was between 97 and 70 Ra units. In addition, a study was conducted to compare the difference in tristimulus values calculation by summation method at 5nm wavelength intervals and tristimulus weighting factors methods at lOnm. This study used the measured relative spectral power distributions of D65 simulators in the visible wavelength range. The influence of various D65 simulators on visual colour matching was investigated by conducting two phases (I and II) visual assessments. The Davidson and Hemmendinger (D & H) Color Rule was used in phase I of visual assessment to test the similarity of three D65 simulators M, V and P by means of colorimetric match point and visual match point. The average visual match point for each D65 simulator was also determined. Fifty normal colour vision observers participated in this visual assessment. The visual match points obtained under three D65 simulators were shown in graphical presentation respectively. The total scores of visual match points were quantified and were further examined by pair comparison t-tests. From the average visual match point results, two D65 simulators M and V had the same average visual match point, one of which has a MI(vis) of 0.24 (Category A) and the other had MI(vis) of 0.34 (Category B) respectively. However, a different average visual match point was observed using D65 simulator P (Category C). In addition, no statistically significant differences were found in pair comparison t-test between D65 simulators M and V. However, it was noticed that there was an uncertainty in total score itself. Therefore, the phase II visual assessment was conducted. The phase II visual assessment further investigated the influence of various D65 simulators on visual colour matching using seventy-seven textile mecameric pairs by grey scale rating method. The American Association of Textile Chemists and Colourists (A.ATCC) Grey Scale for Colour Change was used in this study. A penal of thirty observers, who had experience in colour matching, participated in this visual assessment. Ten out of thirty observers repeated the same visual assessments. Totally, 15400 observations were obtained under five various D65 simulators M, V, P, G and S separately. The agreement between instrumental colour difference and visual data were stated when using relative spectral power distributions of D65 simulators. The visual colour difference data were analysed by the performance factor (PF/3) and pair comparison t-test. The visual colour difference data were also given in graphical presentation. The observer accuracy and observer repeatability were also analysed. A good agreement was obtained between instrumental colour difference and visual data when using relative spectral power distribution of D65 simulators in colour difference calculation of metameric pairs. The results showed that two D65 simulators had smaller PF13 units (11.04), one of which has a MI(vis) of 0.24 (Category A) and the other had MI(vis) of 0.34 (Category B). Another two D65 simulators G and S (Category D) also had smaller PF/3 units (13.08). Moreover, the null hypothesis assumptions of the pair comparison t-tests were not rejected in both cases. Based on the results, the CIE No.51.2 categorization method was discussed. The sensitivity of instrumental colour difference to CIE standard illuminant D65 and different D65 simulators, as well as the difference of reflectance factors were studies using seventy-seven textile metameric pairs. The results showed that, for each textile metameric pairs, the instrumental colour difference changed differently in using relative spectral power distributions of CIE standard illuminant D65 and D65 simulators. The frequency of wavelength intersections of those metameric pairs varied from two to seven within the visible wavelength range, and most of them had three wavelength intersections or above. The differences of reflectance factors were various at different wavelength. The average value of absolute reflectance factor differences at each wavelength from 400nm to 700nm at Snm intervals was determined. Larger average values of absolute reflectance factor differences were noticed in the longer visible wavelength region. Based on the definition of reflectance factor, the Generating Synthetic Metameric Pair (GSMP) model was developed for generating a synthetic metamer from a specified reflectance curve in a given illuminant and primaries. The GSMP model was modified from the additive algorithm, which was original used for generating synthetic reflectance curve. The successful generation of synthetic metameric pair in GSMP model depended on the relative spectral power distribution of illuminant, the nature of reflectance factors and the primaries selections. The CIE 1964 colour matching functions were chosen as default primaries in this model. It was because they had smooth curves and sensitivity to human vision. Five hundreds and sixty-one test colours were used to test the validity of GSMP model. Totally, five hundreds and nineteen synthetic metameric pairs were generated and the successful percentage in generation is ninety-two. The generated synthetic metameric pairs had the similar features as practical metameric pairs like frequency of wavelength interactions and differences in reflectance factors. The generated synthetic metameric pairs had three to five wavelength intersections, which were located near at the prime wavelengths. Larger average values of absolute reflectance factor differences were also noticed in the longer visible wavelength region. The largest average value of absolute reflectance factors difference value was located at 700nm wavelength. The model of Quantifying the Quality of Artificial Light Sources for Visual Assessment (QQALSVA) was developed for quantifying the quality of artificial light sources in visual assessment. This was a mathematical model using colorimetric method to quantify the quality between a specific pair of reference and test artificial light sources. The assumption of QQALSVA model was that the spectral power distribution determines a light source to produce certain reflected colours. Therefore, the instrumental colour difference in synthetic metameric pair between a reference and a test artificial light source was due to their spectral power distributions difference. The GSMP model and CIE special metamerism index: change in illuminant were utilized in QQALSVA model. The GSMP model was used to generate synthetic metameric pair according to the reference artificial light source using eight default test colours. The eight default test colours were originally used in determining the colour rendering index in CIE No. 13.3 method. The CIE special metamerism index: change in illurninant was used to calculate the instrumental colour difference in synthetic metameric pair using test artificial source. The quality performance between the reference and the test artificial light sources was quantified by means of the instrumental colour difference in eight synthetic metameric pairs. The overall quality performance was determined by the average value of eight instrumental colour differences. Testing results were illustrated to demonstrate the validity of QQALSVA model including D65 simulators, CIE F7 and fluorescent lamp D65 simulators, as well as CIE F2, F7 and F1 1. The issue of D65 simulator for practical visual colorimetry was discussed including the suitability for colour matching, colour rendering ability and illumination. In the evaluation of D65 simulator for colour matching, it discussed the validity of the CIE No.51.2 method, the influence of spectral composition on visual assessment. An approach for worldwide colour communication using D65 simulator was suggested. The criterion of colour rendering ability in choosing D65 simulator for critical visual assessment was explored. The results showed that the general colour rending index, Ra, was not a good indicator for D65 simulator in critical visual assessment. It was better to consider both Ra and MI(vis) at the same time. The illumination requirement of D65 simulators in the controlled viewing environment was also stated.