CIE Standards for assessing quality of light sources

Transcription

1 CIE Standards for assessing quality of light sources J Schanda University Veszprém, Department for Image Processing and Neurocomputing, Hungary 1. Introduction CIE publishes Standards and Technical Reports that can be Recommendations, Guides or state of the art reports. In the field of colorimetry there are two Standards which are joint publications of ISO and CIE: CIE colorimetric illuminants (1) CIE colorimetric observers (2) The International Lighting Vocabulary (3), a joint publication with IEC * contains the basic terms and definitions used in colorimetry. CIE is currently updating this fundamental publication. Colorimetry, 2 nd edition (4) contains all the definitions of colorimetric terms, discusses CIE colorimetric practice and provides many useful tables to perform colorimetric calculations. CIE Technical Committee TC 1-48 deals at the present moment with updating this publication. The update will not change anything basic in CIE colorimetry. It will mainly get the recommendations in line with the recommendations passed since One such item is colour difference calculation, where a new colour difference equation was introduced in 1994 (CIE94) (5). Another item is the assessment of the quality of daylight simulators (6), a question we would like to discuss in more detail in this paper. Colour rendering is a fundamental issue in characterising light-sources, and it becomes more and more important also in imaging applications. Thus it will be another item of this paper. The CIE publication on colour rendering (7) also needs some updating, and TC 1-33 is currently working on the subject. The paper will discuss items related to this procedure as well. There are a number of further publications relevant to colour and imaging enumerated in Appendix A. In imaging applications the original is seen either as a scene in real life, or as a paper copy, eventually projected on screen or reproduced by a self-luminous display. The reproduction can be again a paper copy, a projection or a soft-copy on the display. The colours seen on such originals or reproductions depend on the illumination due to two reasons: The reflected/transmitted light of the object seen depends on its illumination; this can be calculated by knowing the spectral power distribution of the source. The illuminant influences also the state of adaptation of our visual system and this controls colour perception. * IEC: International Electrotechnical Commission, Geneva; the ILV is part of the IEC Electrotechnical Vocabulary.

2 In the present paper we will not deal with the chromatic adaptation and colour appearance aspects of the illumination, as other papers of the meeting are discussing this effect in detail. We will concentrate on the more direct effect the illumination has on the colours we perceive. 2. Characterisation of the light source In case of a light source it is usual to distinguish between the colour of the source itself, i.e. the colour we perceive if looking at the source or at an achromatic surface illuminated by the source. In the following this will be termed lamp colour. One has to distinguish between lamp colour and the colour the lamp light has on coloured surfaces. This phenomenon is described by colour rendering: Effect of an illuminant on the colour appearance of objects by conscious or subconscious comparison with their colour appearance under a reference illuminant. (8) The primary physical quantity to determine the colorimetric properties of a light source is its spectral power distribution. For constructing a description of colour rendering one needs according to above definition also a reference illuminant and some objects whose colour is evaluated. Figure 1 shows the spectral power distribution of Standard Illuminant D65 and the necessary amount of three spectral lights (450 nm, 550 nm and 650 nm) to get the same chromaticity coordinates as produced by D65. Spectral power distribution rel. power wavelength, nm Figure 1: Spectral power distribution of D65 and of three monochromatic radiations with the same chromaticity co-ordinates, i.e. metameric to D65. In Figure 2 we show a part of the x,y chromaticity diagram, with the chromaticity of D65, the three line source, and a sample illuminated by these two sources. It is evident from Figure 1 and 2

3 Figure 2 that the chromaticity co-ordinates of the lamp colour are not enough to describe the effect the illuminant will have on surface colours (or in case of transparencies the projector lamp will have on the projected colours). The colour rendering index is the measure for describing the effect lamp-light has on the appearance of object-colours y D line x Figure 2: Chromaticity co-ordinates of the two sources (two coinciding signs on the left) whose spectrum is shown in figure 1 and the chromaticity of a reflection sample illuminated by these two sources (two different locations on the right). 3. Colour rendering As already stated in the previous section, colour rendering serves to describe the effect of an illuminant on the colour appearance of objects. It is based on the assumption that some sources render the colours of our environment in a natural form, i.e. the colour of objects looks under these sources natural, while others distort the rendition. The most fundamental natural source is obviously daylight, thus the primary reference for comparing the rendering of a lamp should be the Standard Illuminant D65. Colour rendering was not an issue in the days before electrical lighting, not even in the early days of incandescent lamp lighting, although it is quite obvious that a low temperature incandescent lamp, as used at the beginning of our century, rendered colours quite differently from daylight. The question of colour rendering became apparent only when gas discharge lamps became available, and lamps with the same lamp colour but different spectra got onto the market. These lamps rendered the colours of objects differently. Due to the fact that incandescent lamps were regarded to render the colours correctly, a description of colour rendering was looked for, where the test lamp could be compared either with a daylight chromaticity or an incandescent lamp chromaticity. It was arbitrarily decided to take as reference illuminant a Planckian-radiator (black-body) if the test lamp had a correlated colour temperature (CCT) below 5000 K, and a phase of daylight if the CCT of the test lamp was higher. In both cases the reference illuminant had to have the same, or nearly the same CCT as the test lamp. The definition of colour rendering stated further on that the colour of objects had to be compared. To do that in a reproducible form, some colour samples (test colour samples, TCS) had to be selected for which the effect of illumination with the reference and test illuminant could be compared. 3

4 3.1 CIE colour rendering index calculation The CIE published its first recommendation on colour rendering index (CRI) calculation in The method now in use was accepted in 1974 (7). This method defines eight basic test colour samples and five supplementary ones, selected from the Munsell book of colour. Table 1 shows the approximate Munsell notation and the colour appearance under daylight. The spectral radiance factors can be found in the CIE recommendation (7) and as a file in (9). k=tcs no. Approximate Munsell notation Colour appearance under daylight (10) 1 7,5 R 6 / 4 Light greyish red 2 5 Y 6/4 Dark greyish yellow 3 5 GY 6/8 Strong yellow green 4 2,5 G 6/6 Moderate yellowish green 5 10 BG 6/8 Light bluish green 6 5 PB 6/8 Light blue 7 2,5 P 6/8 Light violet 8 10 P 6/8 Light reddish purple 9 4,5 R 4/13 Strong red 10 5 Y 8/10 Strong yellow 11 4,5 G 5/8 Strong green 12 3 PB 3/11 Strong blue 13 5 YR 8/4 Light yellowish pink (human complexion) 14 5 GY 4/4 Moderate olive green (leaf green) Table 1: CIE test colour samples used for colour rendering index calculation Reference illuminant, determination of the correlated colour temperature of the test lamp The first step to calculate a colour rendering index is to determine the CCT of the test lamp. This is done using the CIE 1960 Uniform Chromaticity Scale (u,v) diagram**. In this diagram one plots the locus of the Planckian radiators of different temperatures, and determines for which Planckian radiator the chromaticity difference between the test lamp chromaticity and the Planckian locus is a minimum. Stated differently, one draws a perpendicular line from the test lamp chromaticity onto the Planckian locus. This line is called iso-temperature line, as all sources lying on this line have the same CCT. Where this line intersects the Planckian locus, there is the Planckian radiator with the temperature one is looking for. Figure 3 shows in the CIE x,y diagram the Planckian locus and several iso-temperature lines Colour differences between TCS colours illuminated by the two illuminants, effect of chromatic adaptation Once the reference illuminant has been determined, the task is to compare the colour of the TCSs illuminated by the test lamp and the reference illuminant. To do that, one has to determine the tristimulus values of the TCSs, once illuminated by the test lamp and then by the reference illuminant. In the following the symbol T i will be used for the ** The u,v chromaticity co-ordinates are calculated by the following equations: u = 4 x /( 2x+ 12y + 3), v = 6y /( 2x + 12y + 3) 4

5 tristimulus values, with i=1 for X, 2 for Y and 3 for Z. Index t stands for the test lamp, index r for the reference illuminant. The TCS number is shown by the index k (k =0, 1 14, with 0 for the test lamp/reference illuminant). Thus the two sets of tristimulus values are: T i,k,r and T i,k,t The chromaticity difference between T i,k=0,r and T i,k=0,t, i.e. between the two source chromaticities (called in the CIE document adaptive colour shift ) is bridged by a von Kries chromatic adaptation transformation (11). The colour difference between T i,k,r and T i,k,t for k = 1, 14 is then calculated using the CIE 1964 Uniform Colour Space: E i,k,r-t. u,v Iso-Temperature lines y x Figure 3: Planckian locus and some iso-temperature lines calculated using the u,v diagram Calculation of the special and general colour rendering indices The E i,k,r-t values are obviously zero if the test lamp spectrum equals the spectrum of the reference illuminant. Differences between the two spectra will cause differences between the tristimulus values of the TCSs illuminated by the test lamp and the reference illuminant, and thus will produce E i,k,r-t > 0 values; i.e. with increasing difference between the colour rendering properties of the two sources, increasing E i,k,r-t values will be observed. The TC responsible for the development of the colour rendering test method thought that this is less convenient for general application. They suggested a further transformation, so that lamps whose spectrum equals that of the corresponding reference illuminant should have an index of 100, and lamps whose colour rendering deviates from the colour rendering of the reference illuminant should have an index smaller than 100. It was arbitrarily decided that the so called Warm White fluorescent lamp should have a general colour rendering index (see later) of 50. With this in mind, the special colour rendering indices are calculated from the colour differences as: 5

6 R k,t = 100 4,6 E i,k,r-t. To facilitate the evaluation of the colour rendering of a lamp, a further simplification has been introduced: The general colour rendering index (R a ) is the average of the first eight special colour rendering indices: 8 1 R a = R i 8 It has to be emphasised, however, that such an averaging necessarily decreases the information contained in the index. If two lamps have the same general colour rendering index, it does not mean that when illuminating a particular test colour sample with the two sources the colour appearance of the sample will be the same. The comparison of the special colour rendering indices will provide better insight into the similarity of the two lamps, but even in this case the same R i s do not mean that in a complex scene colours will look the same if illuminated with one or the other source. This is even more so if colours produced in different media are compared: e.g. an original interior scene illuminated with a tri-band fluorescent lamp and its reproduction on a VDU screen. There is no standardised method to compare such situations. It is also extremely difficult, as differences between the viewing situations (adaptation, size, etc.) have a complex influence on the colour sensation. 3.2 Possibilities of improving the CRI calculation method As seen in the above section, the currently used CRI method is based at several points on outdated calculation methods. The use of the u,v-diagram to determine iso-temperature lines was a natural choice in the 1960 s, but it can be questioned since the introduction of the CIELUV and CIELAB spaces (4) in We will discuss this issue at the CIM 98 conference (12). The 1964 UCS has also been superseded by the CIELUV and CIELAB spaces. CIE has tentatively adopted a chromatic adaptation transformation (13) and is at present in the course of providing a further recommendation as part of its colour appearance model. We have analysed all these questions in CIE TC 1-33, together with the possibility of using test colour samples readily available, and other methods of selecting reference illuminants (14). This method has not been accepted by the CIE yet, there are some questions still under consideration, but it is hoped that in a not too distant future the colour rendering calculation can be updated and brought in line with present day colorimetric practice. 4. Assessing daylight simulators Besides CRI determination, another method is also in use for evaluating daylight simulators. As well-known with the standardisation of D55, D65 and D75 illuminants, no corresponding sources were standardised. The CIE Technical Report 51 (6) was designed to evaluate how well a simulator (a properly filtered real source) approximates the standard illuminant. Daylight illuminants are used in textile, paper, etc. colorimetry to evaluate fluorescent materials, too, thus this method had to test also the ultraviolet part of the spectral power distribution. It is based on determining metamerism indices. i= 1 6

7 Five sets of metameric data are recommended for testing the simulator. Each set includes the spectral reflection radiance factor of a standard and three metamers (for D55, D65 and D75, resp.) to the standard. The CIE 1964 Supplementary Colorimetric Observer (10 Observer) is used in the calculations. The visible range evaluation is done by calculating tristimulus values for the five metameric pairs (the standard and the metameric pair appropriate for the standard illuminant in question). CIELUV or CIELAB colour difference formulae are used to calculate E i values (i stands for describing the five sets). The visible range metamerism index is calculated from the individual colour differences by: MI vis = 5 i= 1 E i 5 For the ultraviolet range the calculation is a little bit more complicated as the spectral total radiance factor for the fluorescent sample part of each metameric pair is calculated by adding the reflected and emitted parts of the radiation. For that the report provides data of the spectral external radiant efficiency, the spectral reflection radiance factor and the relative spectral distribution of radiance emitted by fluorescence for the three samples. The CIE provides a disk that contains a program to calculate the MI vis and MI uv of a D65 simulator (15). A CIE Technical Committee is discussing the updating of the report that it should include a method and samples to determine a metamerism index also for a D50 simulator which is important in imaging technologies. Table 2 shows the five categories into which sources may be categorised. The MI values are presented both for CIELAB and CIELUV colour difference calculations, as the average length of colour difference vectors in the two colour spaces differ. Category MI (CIELAB) MI (CIELUV) A <0,25 <0,32 B 0,25 < MI <0,5 0,32 < MI <0,65 C 0,5 < MI <1,0 0,65 < MI < 1,3 D 1,0 < MI < 2,0 1,3 < MI < 2,6 E MI > 2,0 MI > 2,6 Table 2: Categories of sources based on metamterism indices. In stating the quality of a daylight simulator by this method, first the category designator for the visible range metamerism index has to be given, followed by the ultraviolet range category descriptor, and the colour space used in brackets. Thus e.g. a daylight simulator of category BC (CIELAB) is one with visible range metamerism index between 0,25 and 0,5 and ultraviolet range metamerism index between 0,5 and 1,0. Such a simulator is found to be useful for many applications. It is expected that when the method becomes updated for D50 simulators, it will provide a useful amendment to the colour rendering index in evaluating e.g. colour matching booth lighting. The CRI evaluates only the visible part of the spectrum, the MI has two parts evaluating both the visible and the ultraviolet range. The MI is also an average value, similarly to the CRI, thus the colour impression a scene provides may be different if illuminated by two simulators whose visible range metamerism index falls into the same category. 7

8 5. CIE Technical Committees working on the above questions During the discussion of light source colorimetric quality evaluation, we have already mentioned that some of the standards and reports are under updating. There are a number of CIE Technical Committees (TCs) working on such updates. Appendix B shows the list of those TCs whose work is most relevant to the questions discussed at this meeting. From this list we would like to draw the attention to the following: TC 1-27 and 1-34 deal with different aspects of colour appearance. At the time of writing this report, the simplified model of colour appearance description, CIECAMs, is in the balloting phase, and it will certainly be published in the very near future. TC 1-21, 1-26, 1-30, 1-42 are extending our knowledge on the brightness sensation. One can hope that based on their work the discrepancy between results predicted by the photometric system and some brightness perceptions will be dissolved. TC 1-36 has the ambitious task to re-evaluate the colour matching functions and provide a more detailed set of such functions to be used also at smaller than 2 angular subtense. This is very important in VDU applications, as colours seen on the screen can be narrow lines. For such colours the 2 Observer is not valid any more. TC 1-38 is looking into the problem of getting colorimetric data into a compatible format, i.e. using them in different applications the same wavelength ranges and increments should apply. The development of practical daylight sources (TC 1-44) and the extension of the index of metamerism to D50 simulators (TC 1-45) progresses as well, although the design of a practical daylight source seems to be a difficult question. The geometric tolerances for colorimetry (TC 2-39) was a subject requested by the imaging technologies, but it turned out that it involves both instrumental and visual investigations and needs detailed review. 6. Summary Light source colorimetry is a subject that is as old as CIE colorimetry. The first CIE recommendations on colorimetry (1931) were passed to help describe light sources and signal lights more completely. During the past 67 years many new recommendations were added to the frame-work of colorimetry, which have profound influence also on light source colorimetry. Most important among them are the introduction of uniform colour scales and colour spaces, especially that of the CIELUV and CIELAB colour spaces, the establishment of new descriptions of colour differences (CIE94 colour difference equation), and most recently the introduction of the CIECAMs colour appearance model. All these new achievements have the consequence that recommendations on light source colour rendering, metamerism indices and similar descriptors of sources have to be reconsidered and brought in line with findings of modern colorimetry. The ever increasing importance of imaging technology in our everyday lives challenges also CIE activity. The establishment of a new CIE Division, to serve the needs of the imaging 8

9 technologies, and especially the use of colour in imaging, is under way. It will be the task of this new CIE Division to co-ordinate research among the different schools working in the field; to organise meetings, seminars and workshops to foster progress; to develop recommendations based on the new research results; and to prepare harmonised standards taking the needs of the different media and industries into account. In summary one can state that CIE Technical Committees and Divisions, working towards the understanding and standardisation of light source and multimedia colorimetry can expect bright and colourful future. References 1 CIE Standard; CIE Colorimetric illuminants, ISO/CIE (CIE S ) 2 CIE Standard; CIE colorimetric observers, ISO/CIE (CIE S ) 3 CIE International Lighting Vocabulary, CIE CIE Technical Report; Colorimetry, 2 nd Edition, CIE CIE Technical Report; Industrial colour-difference evaluation, CIE CIE Technical Report; A method for assessing the quality of daylight simulators for colorimetry, CIE CIE Technical Report; Method of measuring and specifying colour rendering properties of light sources, CIE CIE International Lighting Vocabulary, item , CIE CIE disk, CIE colorimetry and colour rendering tables, CIE D National Bureau of Standards; The ISCC-NBS method of designating colours and a dictionary of colour names, Washington, NBS Circular No. 553, Münch W, Schultz U; Zur Berücksichtigung der Farbumstimmung in der Bewertung der Farbwiedergabe, Farbe, vol. 19, pp Schanda J and co-workers, Effects of white point definition on computer screens, Colour Imaging in Multimedia CIM 98 Conference CIE Technical Report; A method of predicting corresponding colours under different chromatic and illuminance adaptations, CIE Schanda J; The new CIE colour rendering formula, Proc AIC COLOR 97, Vol. 1 pp CIE disk, A method for assessing the quality of D65 daylight simulators for colorimetry (based on Publ. 51), CIE D005 Appendix A CIE Publication relevant to colorimetry and imaging Publ. Nr. Standards ISO/CIE E ISO/CIE E Full title International Standard ISO/CIE 10526: CIE standard colorimetric illuminants International Standard ISO/CIE 10527: CIE standard colorimetric observers 9

10 Techn. Reports Method of measuring and specifying colour rendering properties of light sources Colorimetry, 2nd ed The basis of physical photometry, 2nd ed. 038 Radiometric and photometric characteristics of materials and their measurement 041 Light as a true visual quantity: Principles of measurement 044 Absolute methods for reflection measurements 046 A review of publications on properties and reflection values of material reflection standards 051 A method for assessing the quality of daylight simulators for colorimetry 053 Methods of characterizing the performance of radiometers and photometers 054 Retroreflection: Definition and measurement 059 Polarization: Definitions and nomenclature, instrument polarization 060 Vision and the visual display unit work station 063 The spectroradiometric measurement of light sources 064 Determination of the spectral responsivity of optical radiation detectors 065 Electrically calibrated thermal detectors of optical radiation (absolute radiometers) 069 Methods of characterizing illuminance meters and luminance meters: Performance, characteristics and specifications 075 Spectral luminous efficiency functions based upon brightness matching for monochromatic point sources, 2 and 10 fields 076 Intercomparison on measurement of (total) spectral radiance factor of luminescent specimens 078 Brightness-luminance relations: Classified bibliography 080 Special metamerism index: Change in observer 081 Mesopic photometry: History, special problems and practical solutions 085 Solar spectral irradiance 086 CIE spectral luminous efficiency function for photopic vision 087 Colorimetry of self-luminous displays - A bibliography 095 Contrast and visibility 101 Parametric effects in colour-difference evaluation 103/1 Colour appearance analysis 105 Spectroradiometry of pulsed optical radiation sources 109 A method of predicting corresponding colours under different chromatic and illuminance adaptations 114/1 Survey of reference materials for testing the performance of spectrophotometers and colorimeters (reprint) 114/4 Distribution temperature and ratio temperature 114/5 Terminology relating to non-selective detectors 116 Industrial colour-difference evaluation 118 CIE Collection in Colour and Vision 122 The Relationship between Digital and Colorimetric Data for Computer- Controlled CRT Displays 124 CIE Collection in colour and vision

11 CIE Disks D001 D002 D005 D007 Disc version of CIE photometric and colorimetric data (Publ. 18.2, 86, S001 and S002 tables) CIE colorimetry and colour rendering tables A method for assessing the quality of D65 daylight simulators for colorimetry (based on Publ. 51) A method of predicting corresponding colours under different chromatic and illuminance adaptations (Computer program to Publ. CIE 109) Appnedix B CIE TCs working on items related to colour and imaging No. TITLE 1-21 Testing of Supplementary Systems of Photometry 1-26 Individual Variation of Heterochromatic Brightness Matching 1-27 Specification of Colour Appearance for Reflective Media and Self-Luminous Display Comparisons 1-30 Luminous Efficiency Functions 1-33 Colour Rendering 1-34 Testing of Colour Appearance Models 1-36 Fundamental Chromaticity Diagram with Physiologically Significant Axes 1-38 Compatibility of Tabular Spectral Data for Computational Purposes 1-42 Colour Appearance in Peripheral Vision 1-43 Rod Intrusion in Metameric Colour Matches 1-44 Practical Daylight Sources for Colorimetry 1-45 Revision of CIE Publication 51 to Include D50 Simulators 1-46 Concept and Application of Equivalent Luminance 1-47 Hue and lightness dependent correction to industrial colour difference evaluation 1-48 Revision of CIE Document 15.2 Colorimetry 2-25 Calibration Methods and Photoluminescent Standard for Total Radiance Factor Measurement 2-33 Re-formulation of CIE Standard Illuminants A and D65 (Revision of CIE/ISO 10526) 2-39 Geometric Tolerances for Colorimetry 2-42 The Colorimetry of Visual Displays 11