Table Of Content1
Visual measures of colour
A. K. ROY CHOUDHURY , G ovt. College of Engineering and
Textile Technology, Serampore, India
DOI : 10.1533/9781782423881.1
Abstract : Instrumental colour parameters are very useful for quality
control and colour matching purposes. However, they have poor
correlation with visual parameters of colour. Visual colour order
systems, or colour notations, are very useful for effective communication,
comparison, recording and formulation of colours. The chapter discusses
how colours are assessed visually. Visual perception is a psychological
phenomenon which is diffi cult to measure directly. Hence, various sets of
visual colour parameters are proposed by colourists that are not mutually
convertible.
Key words : colour naming, colour order systems, visual colour attributes,
colour atlas, Pantone, Colour Harmony Manual.
1.1 Introduction
‘ Artists can colour the sky red because they know it’s blue. Those of us who
aren’t artists must colour things the way they really are or people might
think we’re stupid.’ – Jules Feiffer.
Colour is the visual perceptual property corresponding in humans to the
categories red, green, blue and others. Colour derives from the spectrum of
light (distribution of light power versus wavelength) interacting in the eye
with the spectral sensitivities of the light receptors. Colour categories and
physical specifi cations of colour are also associated with objects, materials,
light sources, etc., based on their physical properties such as light absorp-
tion, refl ection and emission spectra. By defi ning a colour space, colours can
be identifi ed numerically by their coordinates. Colour is the element that is
produced when light, striking an object, is refl ected back to the eye.
B erlin and Kay (1 969) described a pattern in naming ‘basic’ colours (such
as ‘red’, but not ‘red-orange’ or ‘dark red’ or ‘blood red’ which are ‘shades’
of red). The authors theorized that as languages evolve, they acquire new
basic colour terms in a strict chronological sequence; if a basic colour term is
found in a language, then the colours of all earlier stages should also be pres-
ent. All languages that have two ‘basic’ colour names distinguish dark/cool
1
© 2015 Elsevier Ltd
2 Principles of colour appearance and measurement
colours from bright/warm colours. The next colours to be distinguished are
usually red and then yellow or green. All languages with six ‘basic’ colours
include black, white, red, green, blue and yellow. The pattern holds up to
a set of 12: black, grey, white, pink, red, orange, yellow, green, blue, purple,
brown and azure (the colour of the sky on a bright, clear day – the hue
halfway between blue and cyan). The work achieved widespread infl uence.
However, the constraints in colour term ordering have been substantially
loosened, both by Berlin and Kay in later publications, and by various crit-
ics. Barbara Saunders (2000) questioned the methodologies of data collec-
tion and the cultural assumptions underpinning the research.
T he colour names always seem to appear in a specifi c order of impor-
tance across cultures – black, white, red, green, yellow and blue. ‘If a popu-
lation has a name for red, it also has a name for black and for white; or, if it
has a name for green, it also has a name for red,’ said researcher Francesca
Tria, a physicist at the ISI Foundation in Turin, Italy. But if a population
has a name for black and white, that does not necessarily mean they have
a name for red. To solve the puzzle of this colour name hierarchy, Tria and
her colleagues devised a computer simulation with pairs of virtual people,
or ‘agents’, who lacked the knowledge of names for colours. One agent,
the speaker, is shown two or more objects, invents a name for a colour
to describe one of the objects, and refers to the item by that colour. The
other agent, the hearer, then has to guess which item, and thus colour, the
speaker referred to. Scientists repeated this until all the agents came to a
consensus on colour names. A key feature of this simulation was its adher-
ence to the limits of human vision. Our eyes are more sensitive to some
wavelengths of light, or colours, than others. The agents in the simulation
were not required to distinguish between hues that a human eye could not
tell apart. ‘Roughly speaking, human eyes can tell apart two colours only if
their wavelengths differ at least by a certain amount – the ‘just noticeable
difference’, Tria said.
T he researchers found that the time agents needed to reach consensus
on a colour name fell into a distinct hierarchy – red, magenta-red, violet,
green-yellow, blue, orange and cyan, in that order. This hierarchy approxi-
mately matches the colour name order seen in real cultures. This hierar-
chy of colours also matches the limits of human vision, with the human eye
being more sensitive to red wavelengths than those for blue, and so on.
‘ Our approach suggests a possible route to the emergence of hierarchi-
cal colour categories’ Tria told Live Science. ‘Humans tend to react most
saliently to certain parts of the spectrum, often selecting exemplars for them,
and fi nally come the process of linguistic colour naming, which adheres to
universal patterns resulting in a neat hierarchy.’
Tria and her colleagues detailed their fi ndings online in the Proceedings
of the National Academy of Sciences (Choi, 2012).
Visual measures of colour 3
Colour is subjective, since it is generated within the visual cortex. Unlike
the sensations of taste, smell or feeling, colour is not a characteristic of objects,
but of the light that enters our eyes from the objects. Objects are visible or
seen coloured only when light reaches our eyes after interaction with them.
The same object may be seen in different colours when observed under vary-
ing lights. In the absence of light, all colours disappear. The common attribu-
tion of colours as properties of objects is largely a matter of memory and
in most cases those refer to some form of sunlight. Daylight is a mixture of
direct sunlight and the scattered component or skylight. We say that snow is
white, soot black, blood red, because under ordinary conditions of life, the
objects appear to be of these hues. While specifying colour, it is, therefore,
essential, to mention the specifi c nature of illumination and viewing.
1.2 Means of colour communication
I t is not very clear how colour names developed historically. One of the two
prevailing opinions is that people of all societies became aware of different
colours or colour categories and then named them in the same sequence:
white and black, red, green, yellow, blue, brown, purple, pink, orange, grey
(Berlin and Kay, 1969). Others think that all colour names are group cultural
achievements and there is little common thread.
M any colour words are related to materials, such as orange, ultramarine,
olive, malachite green, bottle-green, peanut-green, sea-green, etc. These
common names refer to the colours of various common objects, which can
be quickly recognized and memorized by most people. Some names refl ect
poetic invention, such as Cuban Sand, Ashes of Rose, Blue Fox and so on.
But such colour names are very approximate, unreliable and temporary.
Their meaning also changes with observer, time, place, style, technology,
language, culture, etc.
It is common practice to describe colour in terms of hues, such as red, yel-
low, etc., along with tone or secondary hue, such as greenish, bluish, etc., and
the amount of light refl ected such as dark or pale. However, when we describe
a colour as ‘dark greenish blue’, the description is very inadequate, as there
may be many thousands of such colours. The problem was realized long ago.
The accurate description of colour is essential for communication and for
accurate reproduction of colours across a wide range of products. The colour
of any object is commonly registered or recorded in two ways, namely:
1. Preserving coloured physical samples
2. Recording in terms of common colour names
P hysical samples of paint panels, patches of printing inks, coloured papers,
fabrics, yarns or fi bre, etc. are frequently used in the trade. Collections of
4 Principles of colour appearance and measurement
such colour samples are very useful as examples of colour product if the
number of colours required is fairly limited. A good example of such use is
the dye-manufacturer’s ‘shade cards’. Shade cards carry numerous coloured
objects on specifi c substrates (e.g. piece of paper or various textile materi-
als) along with procedures and names of the colourants to be used. However,
the exemplifi cations are very limited. They are restricted to the specifi c type
of colourant or substrate, and cannot be used for general reference.
I t is common practice to describe colour in terms of hues such as red,
yellow, etc., along with tone or secondary hue such as greenish, bluish, etc.,
and the amount of light refl ected, such as dark or pale. However, when we
describe a colour as ‘dark greenish blue’, the description is very inadequate,
as there may be many thousands of such colours. The problem was realized
long ago (Roy Choudhury, 2000).
Colour dictionaries are created for several purposes:
• Standardized colour names facilitate specifi cation, purchase and use of
coloured goods, markers, etc.
• Companies e.g. Pantone register colours (and names), providing formu-
las for inks, plastics, toners and paints to guarantee uniformity and accu-
racy of colour for their clients’ products. These colour designations are
often just numbers and letters, requiring search through process guides
to fi nd a particular shade.
• The combinatorial colour dictionaries underlying the Munsell, OSA-
UCS and GIA colour scales allow fi eld-workers to encode and com-
municate colour from visual observation. These systems endeavour to
partition their colour spaces into equally distinguishable regions with a
named colour at the centre of each.
• A set of colour names can be used to restrict selection when a spectrum
of colours is not available.
• O n computers, summoning colours by common names relieves the
tedium of adjusting or mousing each colour used.
C olour dictionaries for fi eld work must be small enough to make reasonably
quick determinations. A couple of hundred names seems to be an upper
bound.
L arger dictionaries ranging into the thousands were created for identify-
ing coloured textiles, paints and inks. Common colour names (such as ‘blue’)
are not used alone, but they can be components of names. Such collections
of names are meaningless without their charts or samples. These are referred
to as ‘idiosyncratic’.
Colour name dictionaries are three-dimensional datasets with names.
Large dictionaries, ranging into thousands of colours, are created for identi-
fying paints or inks. Common colour names (e.g. ‘blue’) are not used alone;
Visual measures of colour 5
but they can be components of names. These are referred to as ‘idiosyncratic’.
Large collections of idiosyncratic identifi ers convey little meaning with-
out their charts or samples. The large colour dictionaries available online
do not come with charts or sample cards. Aubrey Jaffer ( 2005 ) describes
a method developed for creating usable colour catalogues. The ten-page
charts described in the article are of the Resene paint colours (Resene
Paints Limited, New Zealand). With over 1300 colours, Resene fi lls a large
volume of the CIELAB space uniformly. The primaries (red, green, blue,
cyan, magenta, yellow, white, black) are absent, as they should be for physi-
cally realizable paints. The ‘Resene RGB Values List’ is an excellent source
for surface colours. Plate I (see colour section between pages 146 and 147)
shows one such sheet – page no. 5 (of the ten-page charts) partitioned and
sorted by a Hilbert space-fi lling curve (Jaffer, 2005 ).
It is a daunting task to arrange hundreds or thousands of patches onto
fi xed size sheets. An alphabetic organization produces a visual jumble; suc-
cessfully mapping colours from a three-dimensional colour space to two-
dimensional sheets of paper while keeping similar colours close essentially
reduces the dimension of that colour space. The straightforward method of
3-into-2 reduction is to slice the space in parallel layers, each holding the
same number of colours. After the colours are partitioned into pages, they
are sorted by a second criterion and laid out in a serpentine pattern on the
sheet, going down the fi rst column, then up the second column, then down
the third column, etc. These sheets may be sliced by the luminance (L of
L*C*h), then sorted by hue (h of L*C*h). The colour patches do not transi-
tion smoothly across the page, and the rightmost rows are hard to distinguish
from each other. Statistical dimension reduction is synonymous with data
clustering. Jaffer (2005) reduced dimension by a process which is not depen-
dent on clustering of actual colour coordinates. A space-fi lling function is a
parameterized function which maps a unit line segment to a curve in the unit
square, cube, etc., which gets arbitrarily close to a given point in the unit cube
as the parameter increases. Moon e t al . ( 2001 ) employed a Hilbert space-
fi lling curve which performed well.
1.2.1 C olour notation
W hile communicating or talking about colour, a language which is under-
standable by both the parties must be followed. A logical scheme for order-
ing and specifying colours on the basis of some clearly defi ned attributes
is known as ‘colour notation system’. The attributes are generally three in
number as our vision is trichromatic, and they constitute the coordinates of
the resultant ‘colour space’. Colour notation systems also encompass ‘colour
order systems’, which typically comprise material standards in the form of
6 Principles of colour appearance and measurement
a colour atlas. Due to constraints of the colourant gamut, the atlases may
depict only a physically realizable subset of a colour order system.
Colour notations can be classifi ed into three categories (Rhodes, 2002 ):
1. Device dependent systems – the most common imaging devices used for
reproducing colour are the computer controlled cathode ray tube (CRT)
displays and the colour printers. The associated colour order system and
colour spaces are hardware-oriented and they lack perceptually based
attributes.
2. Mathematical systems – uniform colour spaces based on mathematical
transformation of International Commission on Illumination (CIE) tris-
timulus values such as CIE 1976 (L*, u*, v*) colour space (CIELUV) and
CIE 1976 (L*, a*, b*) colour space (CIELAB) belong to this category.
3. Systems based on database of aim points – colour order systems existing
principally in physical form, the colour samples of which can be measured
to establish a database of aim points. Using interpolation techniques
among limited available samples, many more colours can be defi ned.
1.2.2 History of visual colour ordering
I t is a diffi cult task to deal with the millions of colours which our eyes can
distinguish. We can feel the problem instantly if we try to describe a colour,
particularly its variation from other colours, from memory, or when we try to
describe a colour to a man at a distance via communication channels (Roy
Choudhury, 1996). The problem was known from ancient days, and several
people have tried to solve the problem in their own way. Nobel laureate W.
Ostwald, American artist A. H. Munsell and many others studied the problem
in greater detail. In colourant production and application industries, colours
are to be communicated, compared, recorded and formulated on a regular
basis. This necessitates systematic classifi cation of colours. The objects can be
classifi ed in various ways in terms of colour. The classifi cation may be based
on visually or instrumentally assessed colour parameters. Various colour
order systems were developed, originally on the basis of visual attributes, but
later supported and modifi ed by instrumental assessment. The main reasons
for the widespread interest of colour order systems are for communication
about colour over distance and time as well as for analysis and defi nition of
the aesthetic relations between colours (H ä rd and Sivik, 1983–4).
H umans with normal colour vision can distinguish some two million
colours when viewed against a mid-grey background, and perhaps double
when the background is widely varied (Kuehni, 2 005) . The orderly and
meaningful arrangement has been a matter of concern for last 2000 years.
A colour system which can meet all the requirements needs to be based on
many years of physical and psychological research and experience.
Visual measures of colour 7
T he history of colour order shows that the inter-relationship between
the various colours is rather complex, and it took two millennia to unravel.
Originally, colour order systems consisted of lists of colours, such as those
by Aristotle or Alberti. The great Greek philosopher, Aristotle, was of the
opinion that colour is generated from the interaction of darkness and light,
and that there are seven simple colours out of which all others are obtained
by mixture. Those are white (pure white), yellow, red, purple, green, blue and
black (pure darkness).
At the beginning of the seventeeth century Forsius fi rst represented the
colours in graphical form. A different style of graphical representation of
colour order was developed by the Belgian Jesuit and scholar Franç o is
d’Aguilon (1567–1617). In his graphic representation d’Aguilon showed tonal
mixtures of the three chromatic simple colours with white and black as well
as intermediates between white and black (a grey scale), with arcs above
the line of simples. Below the line he represented with other arcs the hue
mixtures of the three chromatic simple colours (Kuehni, 2003 ).
The modern concept of colour was founded by Isaac Newton (1704). Until
then, all colour order systems were one dimensional or linear. Newton recog-
nized three colour attributes and drew an incomplete (spectral colours only)
chromatic diagram in the form of spectral colours on the circumference and
white in the centre. The saturation lines were drawn as radial lines from the
white centre to the spectral periphery. Newton was also an alchemist, believ-
ing in universal harmony. In analogy to musical tones, he chose seven hues in
the spectrum: violet, indigo, blue, green, yellow, orange and red (VIBGYOR).
However, the choice of seven is always controversial – repeated tests have
shown about 120 discernible colours in the spectrum (Kuehni, 2005 ).
L eBlon (1756) fi rst described that the mixing of pigment colours and the
mixing colours of light are different phenomena. He stated that all visible
objects can be represented by three colours – yellow, red and blue – and
mixtures of these three colours makes black or all other colours. He named
those as material colours, or those used by painters. He further added that
for a mixture of spectral colours, those proposed by Sir Isaac Newton cannot
produce black but, on the contrary, white. Moreover, purple is perceivable
in object colours only.
The German mapmaker and astronomer Tobias Mayer in 1758 fi rst pro-
posed a three-dimensional double tetrahedron colour order system. A
French silk merchant, Gaspard Gré g oire proposed a three-dimensional
object colour order system based on the perceptual attributes hue, (relative)
chroma and lightness and an atlas with 1350 samples was introduced before
1813 (Kuehni, 2 008a) . Matthias Klotz (1748–1821), a German painter also
proposed a three-dimensional colour order system based on independent
perceptual colour attributes. He proposed a cylindrical colour order sys-
tem that consisted of a well-defi ned lightness scale (Kuehni, 2 008b) . About
8 Principles of colour appearance and measurement
100 years later, a very similar colour order system was introduced by Albert
Munsell, based on intensive scientifi c studies.
Helmholtz proposed a four-dimensional Riemannian colour space with the
help of a linear element, which is diffi cult to defi ne precisely and hence the
conceptualization remained unclear. Recent studies (Leonev and Sokolov,
2008) showed that perceived colours can be represented on a spherical
colour space of unit radius (hyper-sphere) in four-dimensional Riemannian
space. The model devotes a dimension to the stimulus parameter ‘dark-
ness’, recognizing the separate signals conveyed by light and dark neuronal
channels. The advantage claimed that the model defi nes mathematically the
relation between the perception of large colour differences and the physi-
cal characteristics of luminous stimuli more consistently. However, a four-
dimensional space is diffi cult to visualize.
1.3 Colour order systems
A colour order system is a systematic and rational method of arranging all
possible colours or subsets by means of material samples. Once the colours
are arranged systematically they are named according to some descriptive
terms and/or are numbered (Graham, 1985 ).
A technical committee of the International Organisation for
Standardisation, ISO/TC187 (Colour Notations), has defi ned a colour order
system as a set of principles for the ordering and denotation of colours, usu-
ally according to defi ned scales (Slideshare, 2013).
A colour order system is usually exemplifi ed by a set of physical samples,
sometimes known as a colour atlas. This facilitates the communication of
colour, but is not a prerequisite for defi ning a colour order system.
The colour order system determines the number of attributes that must
be considered, each attribute defi ning one dimension of the system. For
example, a one-dimensional system may be adequate in the design of light-
ing systems, where it is sometimes suffi cient to consider only CIE luminance
factor (Y), which is a function of the total refl ectance of each surface within
the volume to be lit.
A colour order system is primarily defi ned by a set of material colour
standards, whereas a colour space is essentially a conceptual arrangement.
Over the years, more than 400 colour order systems have been compiled.
On record, the fi rst colour order was devised by Aristotle about 350 b c. It
was vaguely three dimensional, and white was placed opposite black; red,
however, was placed between black and white, red being the colour of the
sky between the states of night and day. Leonardo da Vinci (1452–1519) is
said to have painted sequences in which closely related colours were placed
near each other. Newton (1642–1727), whose discovery of the nature of
white light may be regarded as having begun the science of colour physics,
Visual measures of colour 9
arranged all the hues in a circle, with complementary hues opposite and
white at its centre. These arrangements were two dimensional, however, and
could not therefore include all colours (Slideshare, 2013).
A colour order system is a set of principles that defi nes:
• An arrangement of colours according to attributes such that the more
similar their attributes, the closer are the colours located in the arrange-
ment; and
• A method of denoting the locations in the arrangement, and hence of
the colours at these locations.
It is also desirable that the samples included in any colour order system are
to be properly specifi ed in terms of any standard colorimetric specifi cation,
the most common being CIE colorimetric system.
The targets of colour order systems (Fairchild, 2006 ) are:
• Continuous and orderly arrangement of colours
• A logical system of denotation
• Perceptually meaningful dimensions
• Embodiment with stable, accurate and precise samples.
Colour specifi ers or atlases are a convenient physical form of any colour
order system. Colour order systems are three dimensional, but atlases are
two dimensional so that they can be presented in the form of book or fl at
form (Lewis and Park, 1989). They have multiple functions such as:
• Stand-alone design tool for colour ideas.
• Quick communication of colour ideas over distance.
• The larger swatches provide master standards.
• Basis for specifying colours during colour formulations and colour ideas.
• Supporting role for instrumental response or visual perception of instru-
mentally measured colours.
An atlas should fulfi l certain criteria, such as:
• T he ideal design should be based on colours uniformly distributed
throughout the colour solid.
• Selection of substrate for atlas is very important. Colours illustrated on
cotton are readily matched on other substrates using an appropriate
class of dyes (Park, 2008 ). To facilitate accurate assessments, however,
some atlases have been prepared on multiple substrates. Moreover, dif-
ferent applications require different colour ranges. Gamut requirements
of the textiles, paint, plastics and ceramics are quite different.
10 Principles of colour appearance and measurement
• The ideal atlas should be highly stable and should have good fastness
properties, particularly to light.
• I t should be simple and easily understandable. The samples are to be
reproducible, and replacement pieces should be available.
• It should be cheap, portable and globally used.
However, no atlas is expected to represent visually millions of colours that
can be detected by our eyes. There is no ideal colour order system, and
hence no ideal atlas.
I t is claimed that the RGB colour space atlas developed in 2011 by New
York-based artist Tauba Auerbach (http://taubaauerbach.com/) is a massive
tome (20.3 × 20.3 × 20.3 cm) containing digital offset prints of every varia-
tion of RGB colour possible. It may be considered as a three-dimensional
version of a Photoshop colour picker.
1.3.1 Selection of colour attributes
It is impossible to make physical replica of millions of colours visible to
us. When we have to cover the whole range of possible colours (a million
or more) with a reasonable number of specimens, say a few thousands, the
specimen must be selected according to a well-defi ned system or plan. It is
of utmost necessity to arrange the colours in a systematic manner to inter-
polate or extrapolate the enormous number of perceivable colours from
that limited number of specimens.
I t is well known that the colours are three dimensional. However, the
dimensions of colour are expressed in various ways in different fi elds. For
systematic arrangements, the dimensions should be independent of each
other. The question is, therefore, what dimensions should be chosen to
arrange colours in a three-dimensional space.
T he most natural and logical approach was illustrated by Judd in his
‘Desert island’ experiment (Billmeyer and Saltzman, 1981). A person sit-
ting idly in a desert island may decide to arrange systematically the large
number of pebbles surrounding him according to colour. Firstly he sepa-
rates coloured i.e. chromatic pebbles from colourless i.e. achromatic peb-
bles. Then he arranges colourless pebbles in sequence of black, dark grey,
medium grey, light grey and white (step 1). This classifi cation is based on a
property called ‘Lightness’ or ‘Value’.
T hen he classifi es chromatic pebbles according to their common colour
names. All surface or object colours may be classifi ed broadly into fi ve prin-
cipal colours or hues, namely red, yellow, green, blue and purple (step 2).
While the fi rst four can be seen as spectral colours, purple is perceivable in
object colours only. Furthermore, the variation of colour in the pebbles may