Colour is a perceptual response to light between 380 and 750nm in wavelength, and the approximate correspondence between wavelength and perceived colour12 is shown in figure 2.613. As colour is a perceptual response, it is a subjective phenomenon.
Non-luminous objects must be illuminated to be seen. The source of illumination and the light reflected from an object will have a spectral power distribution (SPD), and the perceived colour will depend on the spectral power distribution. For example, if the peak of the distribution is towards the longer wavelengths, a colour from the red end of the range will be perceived, with the intensity of the colour being a function of the shape of the distribution.
To generate the perception of a colour, recreating the complete spectral power distribution of the light from an object would suffice. In the same viewing conditions, this would indeed recreate the perceived colour, but fortunately, this is not necessary. Colour is commonly represented by three numbers (e.g. the RGB colour values) – tristimulus values that are the basis for all digital encodings of colour. To understand why this is sufficient, and why the viewing conditions affect the perception of colour, it is necessary to understand, in outline, the human visual system.
The basis of human vision is the interpretation of an image projected onto the retina by a lens in the eye. The retina has two types of light sensing elements: rods and cones. Of the two, the rods will give outputs at illumination levels well below that required for the cones to function, but as the rods all have the same spectral response, their output can only indicate luminance, not colour. Therefore, at low light levels, colour vision is absent.
There are three types of cones, named the L, M and S cones after their ability to react to light at long, medium or short (LMS) wavelengths. The S cones have a peak sensitivity towards the blue end of the spectrum, the L cones towards the red end, and the M cones in between, as is shown in figure 2.7. The differing outputs from the three types of cones provide sufficient data for subsequent processing to enable colour perception.
The LMS cone peak sensitivities correspond approximately to the colours blue (S - 447nm), green (M – 540nm) and orange (L – 577nm). It is not immediately apparent
12
The purple colours are not present in the illustration as purples do not correspond to a single wavelength. Purple is seen when both short and long wavelengths (blues and red) are present simulta- neously.
13 Created using a derivative of Earl Glynn’s “Spectra” program fromhttp://www.efg2.com/Lab/
ScienceAndEngineering/Spectra.htm(accessed May 21, 2010) which credits Dan Bruton’s Color Sci- ence page: http://www.physics.sfasu.edu/astro/color.html(accessed May 21, 2010).
Figure 2.7: LMS cone responses to the varying wavelengths of light. The shorter (lower) wavelengths correspond to blues, the longer to reds. Plotted from Stockman and Sharpe (2000) 2◦ cone fundamentals.
how to reconcile the three cone signals with the four perceptually important colours: red, blue, green and yellow.
2.3.1 The derivation of perceptually important primaries
In 1810, Goethe published a wide-ranging treatise that is commonly known as a “The- ory of Colours”14 on his thoughts and observations about colours and its emotional connotations (von Goethe and Matthaei, 1971; von Goethe, 2006). Included was his “chromatic circle” – an arrangement of colours with the hues arranged perceptually, based on the observation of after-images15:
“the colours diametrically opposed to each other in this diagram are those which reciprocally evoke each other in the eye. Thus, yellow demands pur- ple; orange, blue; red, green; and vice versa: thus again all intermediate
14
“Theory of Colours” is the name given as the title by Charles Westlake in his English translation of Goethe’s work. This appears not to be correct. Gimbel (1993) gives the correct title as The Teachings of Colour. MacEvoy (2008), having retranslated the work, also comments on the erroneous title and notes other shortcomings in the translation, such as its incompleteness, with whole sections missing from the English translation. MacEvoy: “Unfortunately, Goethe’s ambitious project has been rendered incoherent both by the deleted sections and by the English translation title: Farbenlehre simply means ‘chromatics,’ with no ‘theory’ implied (just as Sprachlehre means ‘grammar’ and not ‘theory of speech’). Given Goethe’s sensitivity to language, it is not irrelevant to note that the root meaning of lehre is ‘lesson,’ ‘teaching’ or ‘learning from experience’ ”– MacEvoy (2008). For details on the omitted sections and a critique, see MacEvoy’s extensive web site: http://www.handprint.com(accessed May 21, 2010).
15
An after-image is the image seen after fixating on a highly saturated colour for 30–60 seconds and then looking at a white area. The complementary colour will be seen.
graduations reciprocally evoke each other” – von Goethe (2006), ¶5016.
There had been several prior three-dimensional arrangements of colour, but Goethe gives notes the interpolation of the four primary colours and the separateness of blacks, whites and greys:
“. . . yellow, blue, red, green. They represent the most general idea of colour to the imagination, without reference to any very specific modification. If we were to add two other qualifying terms to each of these four, as thus — red-yellow, and yellow-red, red-blue and blue-red, yellow-green and green- yellow, blue-green and green-blue, we should express the gradations of the chromatic circle with sufficient distinctness; and if we were to add the des- ignations of light and dark, and again define, in some measure, the degree of purity or its opposite by the monosyllables black, white, grey, brown, we should have a tolerably sufficient range of expressions to describe ordinary appearances” – von Goethe (2006), ¶610 & 611.
In this, Goethe produces the first perceptually derived colour arrangement of a hue circle (shown on p41), and notes that black, white and grey (and incorrectly, brown) are different from the other colours.
In 1892, Ewald Hering published his “Opponent Process Colour Theory” (Hering, 1964) in which he theorised the existence of four perceptual primary colours arranged as two complementary axes – a red–green axis and a blue–yellow axis – together with a lightness axis. Hering based this theory on observations that, in language, certain colour pairs are never used together when describing colours. There isn’t a blueish- yellow or a greenish-red, suggesting something special about those pairs. The study of after-images makes the opponency of these pairs quite evident. He also noted that colour vision deficiencies often affected either red–green or blue–yellow discrimination. Hering’s theory was later validated experimentally by Hurvich and Jameson (1957).
Hering theorised that there were differing receptors in the eye responsive to light- dark, red-green, and blue-yellow differences. At the time, this supposition was not well received, but while not strictly accurate, it is now known to be substantially correct: there are three sensing elements, and the axes are as he anticipated, but the opponent signals are derived, not primary.
As illustrated in figure 2.8, the opponent colour signals and a luminance signal are calculated from the outputs of the LMS cones by the ganglia before being sent to the optic nerve and the brain (Fairchild, 1998). This recoding of the information by the ganglia transforms the LMS signals into the red–green, blue–yellow and light– dark percepts posited by Hering. Physiologically, addition and subtraction are effected as excitation and inhibition of neurons: for a detailed discussion of neural opponent processing, see Abramov and Gordon (1994).
16
Short Medium Long + + - + + + + + Luminance Red - Green Yellow - Blue - Cone Signals
Figure 2.8: The outputs from the long, medium and short sensitive cones in the eye are combined by the ganglia to derive lumi- nance and opponent colour signals, redrawn from Fairchild (1998). blue green white black red yellow mid-grey
Figure 2.9: The dimensions resulting from the opponent processing: one for lightness, two for colour.
2.3.2 Colour vision deficiencies (CVD)
There are individuals with deficiencies in their colour vision who, for the most part, function perfectly well within society. The widely-used termcolour blindness is a mis- nomer: monochromatism, the complete inability to distinguish colours, is extremely rare, affecting around 0.005% of the population (Fairchild, 1998). Those usually classed as colour blind have either altered colour perception (anomalous trichromacy) or the in- ability to distinguish between certain colours (dichromacy); the frequency of occurrence is shown in table 2.1.
Medical Term Type Male Female
Monochromacy 0.003% 0.002%
Dichromacy Protanopia 1% 0.02%
Deuteranopia 1.1% 0.01% Tritanopia 0.002% 0.001% Anomalous trichromacy Protanomaly 1% 0.02%
Deuteranomaly 4.9% 0.38%
Tritanomaly ∼0 ∼0
Total ∼8.0% ∼0.4%
Table 2.1: The incidence of colour vision deficiencies in West- ern races, derived from Fairchild (1998).
Anomalous trichromacy is most common, affecting around 6.3% of the population (5.9% male, 0.4% female). In anomalous trichromats, the response curves of the LMS cones differ from the norm, resulting in altered colour perception. The effects can be quite subtle and the impairment may only be detected during a colour vision test. The reason for the predominance of those with CVD being male is the adjacency of red– green photo-pigment genes in the X chromosome. In the case of mixing errors, females
have another copy whereas males do not (Manniesing, 2003).
Dichromacy is less common, affecting ∼2.1% of the population, and occurs when one of the colour receptors is either missing or dysfunctional. This results in the viewer losing the ability to distinguish a range of colours. The most common forms are: protanopia, the loss of the red receptors, which affects ∼1% of the male population; and deuteranopia, the loss of green receptors, which also affects ∼1% of the male population. Both result in the loss of red–green–yellow discrimination. The percentage of females affected for both forms is small (0.03%). Impairment of the blue cones (tritanopia) is very rare, affecting less than∼0.003% of the population.
Overall, the most common colour vision deficiencies cause altered colour perception and are primarily restricted to males, with around 8% of the male population and 0.4% of the female population being affected. The deficiencies alter the perception of colour schemes, but do not impair lightness discrimination. If a subgroup of viewers is being addressed, the design of colour schemes could be tailored to limit the effect of a specific impairment. However, omitting hues that cause difficulty for those with the most common impairments would seriously limit the colours that could be used. Therefore, the system being outlined in this research does not specifically address colour deficient viewers. However, noting that lightness perception is unaffected, it may, by ensuring lightness contrast, create schemes that are usable by colour deficient viewers without limiting the hues used.