Content uploaded by Ján Morovic
Author content
All content in this area was uploaded by Ján Morovic on Sep 25, 2015
Content may be subject to copyright.
Content uploaded by Ján Morovic
Author content
All content in this area was uploaded by Ján Morovic on Sep 25, 2015
Content may be subject to copyright.
283
JOURNAL OF IMAGING SCIENCE AND TECHNOLOGY® • Volume 45, Number 3, May/June 2001
Original manuscript received July 6, 2000
▲ IS&T Member
©2001, IS&T—The Society for Imaging Science and Technology
as it is these, which need to be communicated rather
than some characteristics of stimuli depending only on
the receptors in the eye and not on subsequent process-
ing by the visual system. Hence, color appearance mod-
els link the description of the stimulus to the perceptual
attributes it has when seen in a given environment.
These two elements (device characterization and color
appearance modeling) would be sufficient for color im-
age reproduction if all media could reproduce the same
set of colors and if the reproduction process had no im-
age enhancing intents. As this is often not the case, there
is a need for a way of overcoming any differences, that
might exist between the sets of colors obtainable on dif-
ferent media, i.e., it is necessary to have an algorithm
for mapping between their gamuts (Fig. 2.)—a gamut
mapping algorithm (GMA). Finally, there is also a need
for allowing for the pursuit of different rendering in-
tents (e.g., accuracy, pleasantness) that results in the
inclusion of image enhancement in the color reproduc-
tion transform.
In this context, the focus of this article is the gamut
mapping stage of the color reproduction transform with
the aim of contributing to finding such a solution to the
gamut mapping problem, which will give good results
Introduction
Throughout history the art of simulating real life scenes
or creating new images altogether on paper, leather, can-
vas, walls, wood or other materials was superimposed
on a trial and error approach to color image reproduc-
tion. The situation only began to change in this century
with rapid advances in science and technology, which
resulted in the development of various color imaging
technologies and computing on the one hand and a bet-
ter understanding of color on the other. As understood
today, cross-media color image reproduction is a pro-
cess, which includes up to four core elements: device
characterization color appearance modeling, image en-
hancement and gamut mapping, which can be combined
into a six-stage transform1 for the reproduction of color
images (Fig. 1). This transform is an extension of the
five-stage transform2 where image enhancement and
gamut mapping are combined into a single rendering
intent dependent gamut-mapping algorithm.
In color reproduction, the description of an original
image is initially available in a way specific to the me-
dium in which it is present. To be able to transfer color
information between different media, it is first best to
describe them in some medium-independent way. To this
end, device characterization describes color reproduc-
tion media by relating their device-dependent color
specification to the characteristics of the resulting vi-
sual stimulus in terms of how it excites the eye. How-
ever, this alone is not sufficient for color communication,
as a given excitation of the eye can result in different
color appearances depending on viewing conditions.
Further it is necessary to understand the perceptual
attributes of a color (e.g., lightness, chroma and hue),
The Fundamentals of Gamut Mapping: A Survey
Ján Morovic and M. Ronnier Luo▲
Color & Imaging Institute, Derby, United Kingdom
This article aims to give an survey of the fundamentals of gamut mapping by describing the cross-media color reproduction
context in which it occurs, by giving definitions of terms used in conjunction with it, by describing its aims and by giving an
overview of parameters that influence it. These parameters are primarily the choice of color space used, the category into which
a gamut mapping algorithm belongs and whether the approach is image or medium dependent. A succinct summary is then given
of the principal trends in gamut mapping studies conducted to date whereby this summary is based on a survey of a large
number of publications. Individual reviews of these publications can also be accessed on-line at the web site of the CIE Technical
Committee 8–03 on Gamut Mapping on whose behalf this article was submitted for publication.
Journal of Imaging Science and Technology 45: 283-290 (2001)
Figure 1. Six-stage color reproduction transform.
284 Journal of Imaging Science and Technology® Morovic and Luo
under a wide range of conditions. Such a solution can in
some sense be considered to be universal and could be
adopted as a standard.
The importance of having such a solution, which
gives good results for a wide range of original images
and original and reproduction media combinations is
becoming ever more important as the means for color
reproduction are becoming more and more wide-spread.
This trend implies that color reproduction is no longer
a domain of specially-trained experts but is a facility
required by a large audience. Hence, it is of importance
to provide a transparent and unobtrusive system for
color reproduction and any such system will necessar-
ily have to include gamut mapping as an essential fea-
ture. In this setting the availability of a universally
applicable gamut mapping solution is of paramount
importance as potential users would not have the skills
for choosing among a number of algorithms intended
only for application in specific situations and neither
would having to make such a choice be acceptable.
Furthermore a universal algorithm is also of use in a
professional environment, as it can be used as a de-
fault method, which can be supplemented with propri-
etary gamut mapping algorithms designed for the
reproduction of special images or for image reproduc-
tion with special intents.
The need for such a universally applicable gamut
mapping algorithm has also been recognized by the CIE
(Commission Internationale de l’Éclairage) by setting
up the Technical Committee 8-03 on Gamut Mapping
charged with investigating the issue. Indeed, the au-
thors of this article are its members with the first au-
thor being the TC’s chairman and this article is an
excerpt from the committee’s Survey of Gamut Mapping
Algorithms based primarily on the work of Morovic.1 In
addition to the material published in this article the
survey also includes reviews of individual studies on
gamut mapping, which can be found at the CIE Divi-
sion 8 web site—http://www.colour.org/tc8-03/.
The purpose of this article will then be to provide a
survey of literature on gamut mapping rather than a
detailed critical review of individual studies. To have a
better understanding of the state-of-the-art of gamut
mapping and hence the context in which a universally
applicable gamut mapping algorithm is to be sought,
the following areas will therefore be covered: terminol-
ogy, aims of gamut mapping, methods for finding gamut
boundaries, parameters determining the performance
of gamut mapping algorithms and the most prevalent
trends in gamut mapping studies conducted to date.
Terminology
As with any subject, there is a range of possible inter-
pretations of the basic terms used in gamut mapping as
well. Therefore, to avoid misunderstandings, the defi-
nitions used by the CIE TC 8-03 on Gamut Mapping
will be given next, whereby the definitions of an image
were given by Braun and co-workers,3 those of accuracy
and pleasantness by Morovic and Luo4 and the others
by Morovic and Luo:5
Image: two-dimensional stimulus containing pictorial
or graphical information whereby the original image is
the image to which its reproductions are compared in
terms of some characteristic (e.g., accuracy).
Color Reproduction Medium: a medium for display-
ing or capturing color information, e.g., a CRT monitor,
a digital camera or a scanner. Note, that in the case of
printing, the color reproduction medium is not the
printer but the combination of printer, colorants and
substrate.
Color Gamut: a range of colors achievable on a given
color reproduction medium (or present in an image on
that medium) under a given set of viewing conditions—
it is a volume in color space (Fig. 2.).
Color Gamut Boundary: a surface determined by a
color gamut’s extremes.
Gamut Boundary Descriptor (GBD): an overall way
of approximately describing a gamut boundary.
Line Gamut Boundary (LGB): the points of intersec-
tions between a gamut boundary (as characterized by a
GBD) and a given line along which mapping is to be
carried out.
Color Gamut Mapping: a method for assigning colors
from the reproduction medium to colors from the origi-
nal medium or image (i.e., a mapping in color space).
Color Reproduction Intent: the desired relationship
between color information in original and reproduction
media. As a number of solutions to cross-media repro-
duction problems are possible, various color reproduction
intents can be pursued by gamut mapping. The most ge-
neric ones of these are accuracy and pleasantness but it
is also possible to define others for specific applications
(e.g., to provide an accurate reproduction of corporate
identity colors while giving pleasant results for others).
Accurate Reproduction Intent: aims to maximize the
degree of similarity between the original image and a
reproduction of it as is possible given the constraints of
the color reproduction media involved. Note, that the
characteristic of accuracy is intrinsically relative (i.e.,
reproduction versus original)
Pleasant Reproduction Intent: aims to maximize the
reproduction’s correspondence with preconceived ideas
of how a given image should look according to an indi-
vidual whereby this criterion encompasses contrast, lack
of artifacts, sharpness, etc. Note, that unlike accuracy,
pleasantness is absolute—at least as far as a given ob-
server understands it at a given moment.
Figure 2. Color gamuts of a printed medium (solid) and a moni-
tor (transparent) in CIELAB.
The Fundamentals of Gamut Mapping: A Survey Vol. 45, No. 3, May/June 2001 285
Gamut Mapping Aims
On the highest level, the aim of gamut mapping “is to
ensure a good correspondence of overall color appear-
ance between the original and the reproduction by com-
pensating for the mismatch in the size, shape and
location between the original and reproduction gamuts.”1
As a number of colors are physically not reproducible,
it is more advisable to aim for a match of the image’s
appearance rather than the appearance of individual
colors in the image, since the latter can be impossible
for some original colors.
One of the difficulties with implementing this aim is
that there is as yet no model for quantifying the ap-
pearance of complex images and neither is there one for
quantifying the difference between them (though S-
CIELAB6 is a step in this direction). In the absence of
such a model and of an understanding of whether it could
be used effectively in conjunction with gamut mapping,
a number of objectives were heuristically arrived at in
the past and the following aims were identified by
MacDonald2 to be common to the majority of gamut
mapping studies:
•Preserve gray axis of the image and aim for maximum
luminance contrast. This means a mapping of the
original image’s white and black points onto the
reproduction’s white and black points respectively.
•Reduce the number of out-of-gamut colors. Ideally all
the image’s colors should be brought within the
reproduction’s gamut, however, the exclusion of some
extremes is sometimes thought to improve the overall
appearance match. This exclusion would mean the
use of clipping for the excluded extremes and another
method for the remaining colors.
•Minimize hue shifts. It is often thought that when
colors are reproduced the hue of colors needs to be
left unmodified.
•Increase in saturation is preferred. As the
reproduction gamut is already limited in terms of
saturation, at least the available potential should be
used to enable the preservation of saturation
differences present in the original.
Note that the above list represents assumptions made
by some gamut mapping studies whereby the reasons
for making these assumptions are in most cases based
on experience from traditional color reproduction. Even
though experience from traditional color reproduction
is of great value, its maxims need to be looked at care-
fully when used in an environment which enables far
more control over color attributes than was previously
possible.
A further commonly found aim of gamut mapping was
expressed by Stone and co-workers7 when saying that
“the relationship between the colors present was felt to
be more important than their precise value.” This sug-
gests a move of the what-you-see-is-what-you-get
(WYSIWYG) concept onto another level—i.e., it is ap-
plied to images rather than individual colors and could
therefore be called MetaWYSIWYG.8 In spite of playing
down the importance of a match in the traditional sense,
it is still crucial for a color reproduction system to be
able to reproduce individual colors accurately—even
though some original colors cannot reproduced accu-
rately, their modifications need to be reproduced as such.
Calculating Gamut Boundaries
To fulfill the aim gamut mapping has in a particular
color reproduction system, it is first necessary to know
the gamut boundaries of the original and reproduction
gamuts. An understanding of color reproduction media
gamuts has been considered to be of some importance
for some time and was investigated by many research-
ers.9–14 Knowing the boundaries of the gamuts between
which mapping is to be carried out is essential for the
majority of GMAs developed to date and can be divided
into two separate problems.
First, it is necessary to compute a gamut boundary
descriptor (GBD), i.e., some overall way of approximately
describing a gamut. For media gamuts this can be done
either directly from specific characterization models—
e.g., Kubelka-Munk equations15 or Neugebauer equa-
tions16—or using methods which can be applied to any
characterization model.17 Further there are also some
methods which can be used for computing the gamuts
of images as well as media18,19 and there have also been
some studies that looked more specifically at image
gamut calculation.20,21
Second, it is also important to be able to find the in-
tersections between the gamut boundary (as computed
using the above methods) and a given line along which
mapping is to be carried out—the line gamut boundary
(LGB). The articles published by Herzog,22,23 Braun and
Fairchild24 and Morovic and Luo5,25 describe methods for
doing this as well as obtaining the initial gamut bound-
ary descriptor. Note that the first three of these meth-
ods are aimed at obtaining the LGB for lines of constant
hue and lightness, which are most often used by present
GMAs. The last two of the above articles describe meth-
ods for finding the gamut boundary along lines of con-
stant spherical coordinates and along any line of
constant hue angle respectively. Clearly all these meth-
ods can be used to calculate the gamut boundary along
any line when iterative techniques are employed.
Gamut Mapping Parameters
Color Space
As can be seen from the definitions of a color gamut
and of gamut mapping given earlier, they are both
closely associated with color spaces. Most gamut map-
ping algorithms intend to work with perceptual at-
tributes, i.e., colorfulness, chroma, saturation,
brightness, lightness, hue or color names (e.g., red, dark
green, orange, etc. whereby each of these would repre-
sent a subset of colors represented by a volume in color
space) and to make this possible, they are implemented
in color spaces which predict them. More specifically,
they usually intend to maintain some of a color’s per-
ceptual attributes while changing others. If under these
circumstances the predictors are imperfect, changes in
the predictor of one attribute can also result in changes
of another perceptual attribute (e.g., in some cases, if
the L*—the predictor of lightness in CIELAB—of a color
is changed, its perceived hue or chroma might also
change).
In the color spaces that are most often used for gamut
mapping—CIELAB and CIELUV26 and in some cases
LLAB27 or RLAB28—there are problems especially with
the predictors of hue. In particular there are deficien-
cies in the uniformity of hue angles in the blue region of
CIELAB (e.g., hue angles of around 290°) which can re-
sult in changes of perceived hue when only the L* or C*
of a color is changed. The performance of the hue pre-
dictor in CIECAM97s29 is somewhat better for this re-
gion. More detail on the performance of hue predictors
of various color spaces can be found in articles by Hung
and Berns30 and Ebner and Fairchild.31 Of particular
286 Journal of Imaging Science and Technology® Morovic and Luo
interest are also the articles by Ebner and Fairchild,32
which describes the IPT color space developed so as to
have improved hue uniformity, the article by Marcu33
describing the mLAB space developed for the same rea-
son, the article by McCann34 discussing some shortcom-
ings of CIELAB and the article by Zeng35 proposing a
gamut mapping solution using different color spaces for
different parts of color space.
When implementing or evaluating GMAs, it is impor-
tant to understand the deficiencies of the color space
used for the mapping and not to confuse the color space’s
predictor with the predicted perceptual attribute (e.g.,
in CIELAB hab is not hue and L* is not lightness—they
are only their predictors).
Gamut Mapping Categories
Once the necessary gamuts are known in the chosen
color space, it is possible to implement a GMA. While
there are a wide variety of gamut mapping solutions,
many of them can be categorized in terms of the follow-
ing parameters:
Type of Mapping
Gamut Clipping. Gamut clipping algorithms only
change colors which are outside the reproduction gamut
either from the very beginning or after lightness com-
pression. For colors outside the reproduction gamut,
these algorithms specify a mapping criterion, which is
used for finding the point on the reproduction gamut to
which a given original color is mapped.
Gamut Compression. Gamut compression algorithms
are applied to all colors from the original gamut so as to
distribute the differences caused by gamut mismatch
across the entire range. Compression is needed when
larger differences are to be overcome, as gamut clipping
could result in unacceptable loss of variation in out-of-
gamut regions under such circumstances.
There are a number of ways how gamut compression
can be carried out, whereby it can either be uniform (i.e.,
determined in the same way for all colors) or non-uni-
form (having different parameters depending on the at-
tributes of original colors to which they are applied) and
the following approaches are the most common (Fig. 3.):
• linear compression
• non-linear compression
• piece-wise linear compression
• higher-order polynomial compression
•combination of different compression methods (e.g.,
soft-clipping can be a three-part compression method
whereby no change is applied to part of the range,
clipping to another part and there is a higher-order
transition between the two).
Direction of Mapping. It is not only the type of map-
ping that characterizes a particular gamut mapping ap-
proach but also the choice of lines along which the
mapping is applied. In most cases gamut mapping is
carried out along the following directions (Fig. 4.):
•lines of constant perceptual attribute predictors (e.g.,
constant lightness and hue, constant saturation and
hue)
•lines towards single center-of-gravity (e.g.,
compression towards L* = 50 on lightness axis)
•lines towards variable centers-of-gravity (e.g.,
compression towards lightness of cusp on lightness
axis)
•lines towards the nearest color in the reproduction
gamut (e.g., as is the case with minimum ∆E clipping)
Figure 3. Gamut clipping and compression along a given line
(Ogb and Rgb represent of the original and reproduction gamut
boundary respectively).
Figure 4. Gamut mapping directions.
The Fundamentals of Gamut Mapping: A Survey Vol. 45, No. 3, May/June 2001 287
It is also important to understand the fundamental
interconnectedness of mapping type, mapping direction
and the color space in which gamut mapping is carried
out, as choices in any one of the former two influence
each other and they are both influenced by the latter (e.g.,
hue shifts in the blue region of CIELAB might give good
results due to the color space’s hue non-uniformity25).
Image versus Medium Gamuts
When gamut compression is used, there arises a ques-
tion as to which gamuts to map between. This is the
case as the original gamut can be seen as either the
gamut of the original medium or the gamut of the origi-
nal image (i.e., a subset of the original medium’s gamut
(Fig. 5)). For current gamut clipping methods this is not
an issue as it is sufficient to know the reproduction
gamut to which any irreproducible colors from the origi-
nal are mapped.
To make as few changes to the original image as pos-
sible, it is more reasonable to use the image gamut as
the original gamut since this means that colors are only
modified when necessary. Indeed there are a number of
experimental studies that support the idea that the use
of image gamuts gives preferred reproductions.36–39
If, on the other hand, media gamuts are used, a given
image could be modified to allow for colors which are
not present in it (e.g., when the medium gamut is used
an image’s colors are changed even if all of them are in
the reproduction gamut to begin with). However, there
is a practical advantage to mapping between media gam-
uts as lookup tables (LUTs) can be calculated from them
and then used for transforming an image without know-
ing its individual gamut.
Trends in Gamut Mapping Studies
As already mentioned, the present article is based on a
Survey of Gamut Mapping Algorithms of the CIE TC 8–
03 on Gamut Mapping. That survey includes reviews of
57 individual articles published between 1977 and the
first quarter of 1999 and this section is a summary of
its findings as well as of articles published before the
end of 2000. The reviews showed that a wide variety of
gamut mapping strategies have been proposed and in
some cases also evaluated in the past. What will be at-
tempted here is an identification of the more prevalent
approaches and those that seem to be particularly prom-
ising or inductive of future work.
First, one of the most noticeable trends in the reviewed
gamut mapping work is the agreement among different
studies that image-dependent methods are preferred
over medium-dependent methods, which is in some sense
supported by a number of sources.36–44 At the same time,
however, there is some work that suggests that while
determining gamut mapping on an image-dependent
basis gives better results, it is not an image’s gamut
that determines how it ought to be gamut mapped.20
Second, there is significant number of studies where
clipping is given preference over compression36–38,45–55
whereby this is done implicitly in some cases. In some
of these articles minimum ∆E clipping is used by de-
fault and in others clipping algorithms are proposed
without reference to compression. In addition, there is
a also good number of articles among the above which
have arrived at the preference of clipping by means of
well-designed psychophysical experiments.36–38, 52, 53 The
exception to this trend is the article by Morovic and
Luo,56 where clipping performed significantly worse
than compression. A possible explanation of this is that,
in all the former cases the relationship between origi-
nal and reproduction gamuts was either artificial, rela-
tively small (when compared with gamut differences
between the media in Morovic and Luo56) or there was
no lightness difference between them and therefore it
was sufficient to use clipping. For overcoming larger
gamut differences, however, it seems to be advanta-
geous to use compression. The verification of this hy-
pothesis would be of particular use to arriving at a
universal GMA.
Third, while the vast majority of algorithms published
before 1999 start with uniform overall lightness com-
pression, there are a number of solutions that propose
other mappings. The earliest of these is the work of Ito
and Katoh57 where no initial lightness mapping is ap-
plied and both lightness and chroma differences are
overcome in a single step. The GMAs proposed by
Morovic and Luo5,25,56,58,59 again either have no initial
lightness mapping (for the SLIN or CUSP GMAs) or map
lightness in a chroma-dependent way (GCUSP) whereby
maintaining more of the chroma of more chromatic col-
ors. Braun and Fairchild60–64 then suggest the use of sig-
moidal lightness mapping which gives more weight to
the middle than to the extremes of the lightness range
whereby preserving more overall contrast. Finally Kang
and co-workers65,66 suggest lightness clipping and Braun
and co-workers53 suggest a luminance compression using
an inverted power function in a linear RGB color space
for the purposes of pleasing color reproduction. Which
of these is the best solution again seems to depend on
gamut difference and requires further verification.
Fourth, the preservation of hue (or hue angle) is also
a point which occurs in all but the following
articles25,39,58,67,68 and the articles where minimum ∆E
clipping is used. A related issue is also the preservation
of hue names which the work of Motomura69–71 has par-
ticularly focused on.
Fifth, there are a good number of articles which sug-
gest the use of different mapping methods for different
parts of color space.25,35,39,52,54,55,57,58,67,68,72–75
Sixth, regarding the direction of mapping the ap-
proaches tried most frequently involve mapping along
lines that are constant in some perceptual attributes
(e.g., lightness and hue for the LLIN algorithm1), lines
that converge on a single point (e.g., the point on the
lightness axis having the lightness of the cusp as is the
Figure 5. Gamut of CRT (mesh) and MUS image (solid) in
CIELAB.
288 Journal of Imaging Science and Technology® Morovic and Luo
case for GCUSP1), lines that have the same slope,77 lines
that are determined by smallest color difference,78–80
lines that are curved81,82 or lines that are determined by
corresponding distances along the original and repro-
duction boundaries.83
In terms of gamut clipping, the method proposed by
Katoh and Ito49 seems to be a good solution, not least
because of its simplicity and good correlation with the
results of the experimental study of Ebner and
Fairchild.51 In addition the relative importance of L ≥ h
> C used in this model is also confirmed by Wei et al.39
Recently, Ito and Katoh84 have published a more exten-
sive investigation of minimum color difference gamut
clipping and have recommended the use of ∆E9485 in
CIELUV and ∆EBFD86 in CIELAB. More recently a gamut
clipping solution that aims to preserve spatial luminance
or lightness variation has also been proposed87 and it
too seems to be a promising approach well worth fur-
ther evaluation.
In terms of gamut compression algorithms, there is a
degree of inhomogeneity between the various proposals,
whereby the GCUSP algorithm proposed by Morovic and
Luo25 and the GMA using sigmoidal lightness compres-
sion and proposed by Braun and Fairchild60–64 seem most
promising. Of these GCUSP has been most extensively
tested and found to perform well for various printed me-
dia as well as both in the CIELAB and CIECAM97s color
spaces.59 In addition to its accuracy it has also been found
that reproductions made with it are pleasant when the
original images are pleasant.4 The sigmoidal lightness
compression algorithm, on the other hand has under some
conditions been shown to perform better than GCUSP
and has been tested independently as well.20 However,
due to the variation in results of gamut compression stud-
ies, more work is needed for seeing what algorithm is
best suited for a default solution.
Finally, there are some articles describing interest-
ing approaches, which either rely on a more satisfac-
tory solution of other problems before they can become
effective (e.g., Nakauchi and co-workers,42,43 which uses
an image difference model) or which are meant for spe-
cific applications (e.g., Braun and co-workers,88 which
is specifically for business graphics).
Most of the techniques reviewed here were closely
related to CRT and printed media. In this context, other
methods were also tried to reduce the gamut mismatch
problem by using more than four inks, in which case
the gamut of the printed medium is increased. However,
it was suggested that some of these systems do not ac-
tually provide significant improvements89 and therefore
do not diminish the need for gamut mapping. Attempts
have also been made to obtain exact color matches (i.e.,
relative luminances, CIE chromaticities and absolute
luminances are identical) between CRT and print,90 how-
ever, this means that only colors from the overlap of the
two gamuts can be used. Due to this intensive focus on
CRT to print mapping in the past, it would be very use-
ful to conduct studies that use different media combi-
nations, like transparency to CRT or print, in future.
Gamut mapping is also of importance in other fields
and, as an example, proprietary solutions have been
devised for the ray tracing of prisms and rainbows91
and for the joining together—sometimes referred to as
seamless welding—of several images (e.g., for pan-
oramic pictures).92
Conclusions
Arriving at a universally applicable gamut mapping
algorithm is an aim shared by many researchers and
industrialists working in this area and the purpose of
this article was to make the common basis arrived at
by the members of the CIE TC 8–03 on Gamut Map-
ping more widely available. The terminology, param-
eters influencing gamut mapping and view of past
research presented here are those of the technical com-
mittee and are important prerequisites for effective
cooperation on the development of future, generally
acceptable solutions.
Acknowledgment. The authors would like to thank all
their fellow members of the CIE TC 8–03 on Gamut
Mapping: Gus Braun (Rochester Institute of Technol-
ogy, USA), Bodrogi Péter (University of Veszprém, Hun-
gary), Fritz Ebner (Xerox, USA), Mark D. Fairchild
(Rochester Institute of Technology, USA), Patrick Herzog
(Aachen University of Technology, Germany), Tony
Johnson (UK), Naoya Katoh (Sony, Japan), Marc Mahy
(Agfa, Belgium), Gabriel Marcu (Apple, USA), John
McCann (USA), Ethan Montag (Rochester Institute of
Technology, USA), Hideto Motomura (Rochester Insti-
tute of Technology, USA), Todd Newman (Canon Infor-
mation Systems, USA), Raimondo Schettini (Italian
National Research Council, Italy), Geoff J. Woolfe
(Kodak, USA). Their comments and suggestions regard-
ing the present article were of great importance.
References
1. J. Morovic,
To Develop a Universal Gamut Mapping Algorithm
, Ph.D.
Thesis, University of Derby, Derby, United Kingdom (1998)
2. L. W. MacDonald, Gamut Mapping in Perceptual Color Space, in
Proc. 1
st
IS&T/SID Color Imaging Conf.
, IS&T, Springfield, VA, 193–
196 (1993)
3. K. M. Braun, M. D. Fairchild and P. J. Alessi, Viewing Techniques
for Cross-Media Image Comparisons,
Color Res. Appl.
21, 6–17
(1996)
4. J. Morovic and M. R. Luo, The Pleasantness and Accuracy of Gamut
Mapping Algorithms, in
ICPS Conf. Proc.
, Antwerp, Belgium, 2, 39–
43 (1998)
5. J. Morovic and M. R. Luo, Developing Algorithms for Universal Color
Gamut Mapping,
Color Engineering: Vision and Technology
, L. W.
MacDonald and M. R. Luo, Eds., John Wiley & Sons, Chichester,
UK, 253–283 (1999)
6. X. M. Zhang and B. A. Wandell, A spatial extension to CIELAB for
digital color image reproduction,
SID Symposiums Technical Digest,
27, 731–734 (1996)
7. M. C. Stone, W. B. Cowan and J. C. Beatty, Color Gamut Mapping
and the Printing of Digital Color Images,
ACM Transactions on Graph-
ics
7, 249–292 (1988)
8. J. Morovic, To Achieve WYSIWYG Color via Adobe Photoshop,
Course Notes of Easter School ’96: Color Management for Informa-
tion Systems
, Color and Imaging Institute, Derby, UK (1996)
9. P. Oittinen, H. Autio and H. Saarelma, Color Gamut in Halftone Print-
ing,
J. Imaging Sci. Technol.
36, 496–501 (1992)
10. K. Schläpfer and E. Widmer, Which Color Gamut Can Be Achieved
in Multicolor Printing and in Television?, in
TAGA Proc.
, TAGA, Roch-
ester, NY, 41–49 (1993)
11. J. A. Bristow, M. Klaman and M. Hoc, The Comparison of Color Gam-
uts, in
IARAIGAI Conf. Proc. Advances in Printing Sci. Tech.
, 24,
311–332 (1994)
12. G. Marcu and S. Abe, Ink Jet Printer Gamut Visualisation, in
Proc.
11
th
IS&T Intl. Cong. on Advanced Non-Impact Printing Tech.
, IS&T,
Springfield, VA, 459–462 (1995)
13. G. Marcu and S. Abe, Three-Dimensional Histogram Visualisation
in Different Color Spaces and Applications,
J. Electronic Imaging
4,
330–346 (1995)
14. G. W. Meyer and C. A. Robertson, A Data Flow Approach to Color
Gamut Visualization, in
Proc. 5
th
IS&T/SID Color Imaging Conf.
, IS&T,
Springfield, VA, 209–214 (1997)
15. P. G. Engeldrum, Computing Color Gamuts of Ink-jet Printing Sys-
tems, in
SID Proc.
, SID, San Jose, CA, 27, 25–30 (1986)
16. M. Mahy, Calculation of Color Gamuts Based on the Neugebauer
Model,
Color Res. Appl.
22, 365–374 (1997)
17. M. Inui, Fast Algorithm for Computing Color Gamuts,
Color Res.
Appl.
18, 341–348 (1993)
18. W. Kress and M. Stevens, Derivation of 3-Dimensional Gamut De-
scriptors for Graphic Arts Output Devices, in
TAGA Proc.
, TAGA,
Rochester, NY, 199–214 (1994)
The Fundamentals of Gamut Mapping: A Survey Vol. 45, No. 3, May/June 2001 289
19. T. J. Cholewo and S. Love, Gamut Boundary Determination Using
Alpha-Shapes, in
Proc. 7
th
IS&T/SID Color Imaging Conf.
, IS&T,
Springfield, VA, 200-204 (1999)
20. J. Morovic and P. L. Sun, The Influence of Image Gamuts on Cross-
Media Color Image Reproduction, in
Proc. 8
th
IS&T/SID Color Imag-
ing Conf.
, IS&T, Springfield, VA, 324–329 (2000)
21. R. Saito and H. Kotera, Extraction of Image Gamut Surface and
Calculations of its Volume, in
Proc. 8
th
IS&T/SID Color Imaging Conf.
,
IS&T, Springfield, VA, 330–334 (2000)
22. P. G. Herzog, Analytical Color Gamut Representations,
J. Imaging
Sci. Technol.
40, 516–521 (1996)
23. P. G. Herzog, Further Development of the Analytical Color Gamut
Representations,
SPIE Proc.
, SPIE, Bellingham, WA, 3300, 118–
128 (1998)
24. G. J. Braun and M. D. Fairchild, Techniques for Gamut Surface Defi-
nition and Visualization, in
Proc. 5
th
IS&T/SID Color Imaging Conf.
,
IS&T, Springfield, VA, 147–152 (1997)
25. J. Morovic and M. R. Luo, Gamut Mapping Algorithms Based on
Psychophysical Experiment, in
Proc. 5
th
IS&T/SID Color Imaging
Conf.
, IS&T, Springfield, VA,44–49 (1997)
26. CIE, CIE Publication 15.2,
Colorimetry
, Second Edition (1986)
27. M. R. Luo and J. Morovic, Two Unsolved Issues in Color Manage-
ment—Color Appearance and Gamut Mapping, in
Proc. 5th Interna-
tional Conference on High Technology: Imaging Science and
Technology—Evolution & Promise
, Chiba, Japan, 136–147 (1996)
28. M. D. Fairchild,
Color Appearance Models
, Addison–Wesley, Read-
ing, MA (1998)
29. M. R. Luo and R. W. G. Hunt, The Structure of the CIE 1997 Color
Appearance Model (CIECAM97s),
Color Res. Appl
. 23, 138–146
(1998)
30. P. C. Hung and R. S. Berns, Determination of Constant Hue Loci for
a CRT Gamut and Their Predictions Using Color Appearance Spaces,
Color Res. Appl.
20, 285–295 (1995)
31. F. Ebner and M. D. Fairchild, Finding Constant Hue Surfaces in
Color Space, in
SPIE Proc.
, SPIE, Bellingham, WA, 3300, 107–
117 (1998)
32. F. Ebner and M. D. Fairchild, Development and Testing of a Color
Space (IPT) with Improved Hue Uniformity, in
Proc. 6
th
IS&T/SID
Color Imaging Conf.
, IS&T, Springfield, VA, 8–13 (1998)
33. G. Marcu, Gamut Mapping in Munsell Constant Hue Sections, in
Proc. 6
th
IS&T/SID Color Imaging Conf.
, IS&T, Springfield, VA, 159–
162 (1998)
34. J. J. McCann, Color Gamut Measurements and Mapping: The Role
of Color Spaces, in
SPIE Proc.
, SPIE, Bellingham, WA, 3648, 68–
82 (1999)
35. H. Zeng, Gamut mapping in multiple color spaces, in
SPIE Proc.
,
SPIE, Bellingham, WA, 3963, 301-306 (1999)
36. R. S. Gentile, E. Walowitt and J. P. Allebach, A comparison of tech-
niques for color gamut mismatch compensation,
J. Imaging Technol.
16, 176–181 (1990)
37. E. G. Pariser, An Investigation of Color Gamut Reduction Techniques,
in
IS&T’s 2
nd
Symposium on Electronic Publishing
, IS&T, Spring-
field, VA, 105–107 (1991)
38. E. D. Montag and M. D. Fairchild, Psychophysical Evaluation of
Gamut Mapping Techniques Using Simple Rendered Images and
Artificial Gamut Boundaries,
IEEE Trans. Image Proc.
, 6, 977–989
(1997)
39. R. Y. C. Wei, M. J. Shyu and P. L. Sun, A New Gamut Mapping Ap-
proach Involving Lightness, Chroma and Hue Adjustment, in
TAGA
Proc
., TAGA, Rochester, NY, 685–702 (1997)
40. T. Hoshino and R. S. Berns, Color Gamut Mapping Techniques for
Color Hard Copy Images, in
SPIE Proc.
, SPIE, Bellingham, WA,
1909, 152–164 (1993)
41. L. W. MacDonald and J. Morovic, Assessing the Effects of Gamut
Compression in the Reproduction of Fine Art Paintings, in
Proc. 3
rd
IS&T/SID Color Imaging Conf.
, IS&T, Springfield, VA, 194–200 (1995)
42. S. Nakauchi, M. Imamura and S. Usui, Color Gamut Mapping by
Optimizing Perceptual Image Quality, in
Proc. 4
th
IS&T/SID Color
Imaging Conf.
, IS&T, Springfield, VA, 63–67 (1996)
43. S. Nakauchi, S. Hatanaka and S. Usui, Color gamut mapping based
on a perceptual image difference measure,
Color Res. Appl.
24, 280-
291 (1999)
44. S. D. Kim, C. H. Lee, K. M. Kim, C. S. Lee, and Y. H. Ha, Image-
dependent gamut mapping using a variable anchor point, in
SPIE
Proc.
, SPIE, Bellingham, WA, 3300, 129–137 (1998)
45. J. J. Sara,
The Automated Reproduction of Pictures with
Nonreproducible Colors
, PhD. Thesis, Massachusetts Institute of
Technology, Cambridge, MA, (1984)
46. J. Meyer and B. Barth, Color Gamut Matching for Hard Copy, in
SID
89 Digest
, SID, San Jose, CA, 86–89 (1989)
47. J. M. Taylor, G. M. Murch and McManus Tektronix HVC: A Uniform
Perceptual Color System, in
SID Digest of Tech. Papers,
SID, San
Jose, CA (1989)
48. R. S. Berns and H. K. Choh, Cathode-Ray-Tube to Reflection-Print
Matching Under Mixed Chromatic Adaptation Using RLAB,
J. Elec-
tronic Imaging
4:4, 347–359 (1995)
49. N. Katoh and M. Ito, Gamut Mapping for Computer Generated Im-
ages (II), in
Proc. 5
th
IS&T/SID Color Imaging Conf.
, IS&T, Spring-
field, VA, 126–129 (1996)
50. G. Marcu and S. Abe, Gamut Mapping for Color Simulation on CRT
Devices, in
SPIE Proc.
, SPIE, Bellingham, WA, 2658, 308–315
(1996)
51. F. Ebner and M. D. Fairchild, Gamut Mapping from Below: Finding
the Minimum Perceptual Distances for Colors Outside the Gamut
Volume,
Color Res. Appl.
22, 402–413 (1997)
52. P. G. Herzog and M. Müller, Gamut Mapping Using an Analytical
Color Gamut Representation, in
SPIE Proc.
, SPIE, Bellingham, WA,
3018, 117–128 (1997)
53. K. M. Braun, R. Balasubramanian and R. Eschbach, Development
and Evaluation of Six Gamut-Mapping Algorithms for Pictorial Im-
ages, in
Proc. 7
th
IS&T/SID Color Imaging Conf.
, IS&T, Springfield,
VA, 149–154 (1999)
54. C. Hung-Shing, M. Omamiuda and H. Kotera, Gamut Mapping
Method Adaptive to Hue-Divided Color Distributions, in
Proc. 7
th
IS&T/SID Color Imaging Conf.
, IS&T, Springfield, VA, 289–294 (1999)
55. C. Hung-Shing, M. Omamiuda and H. Kotera, Adaptive Gamut Map-
ping Method Based on Image-to-Device, in
Proc. IS&T’s NIP 15:
International Conference on Digital Printing Technologies
, IS&T,
Springfield, VA, 346-349 (1999).
56. J. Morovic and M. R. Luo, Cross-media Psychophysical Evaluation
of Gamut Mapping Algorithms,
Proc. AIC Color 97 Kyoto
, Kyoto,
Japan, 2, 594–597 (1997)
57. M. Ito and N. Katoh, Gamut Compression for Computer Generated
Images, in
Extended Abstracts of SPSTJ 70th Anniversary Sympo-
sium on Fine Imaging
, SPSTJ, Tokyo, Japan, 85–88 (1995)
58. J. Morovic and M. R. Luo, A Universal Algorithm for Color Gamut
Mapping,
Proc. CIM ’98 Conf.
, Derby, UK (1998)
59. J. Morovic and M. R. Luo, Verification Of Gamut Mapping Algorithms
In CIECAM97s Using Various Printed Media, in
Proc. 6
th
IS&T/SID
Color Imaging Conf.
, IS&T, Springfield, VA, 53–56 (1998)
60. G. J. Braun and M. D. Fairchild, Image Lightness Rescaling Using
Sigmoidal Contrast Enhancement Functions,
J. Electronic Imaging
8:4, 380–393 (1999)
61. G. J. Braun and M. D. Fairchild, General-Purpose Gamut Mapping
Algorithms: Evaluation of Contrast-Preserving Rescaling Functions
for Color Gamut Mapping, in
Proc. 7
th
IS&T/SID Color Imaging Conf.
,
IS&T, Springfield, VA, 167–172 (1999).
62. G. J. Braun,
A Paradigm for Color Gamut Mapping of Pictorial Im-
ages
, PhD. Thesis, Rochester Institute of Technology, Rochester,
NY (1999)
63. G. J. Braun and M. D. Fairchild, General-Purpose Gamut-Mapping
Algorithms: Evaluation of Contrast-Preserving Rescaling Functions
for Color Gamut Mapping,
J. Imaging Sci. Technol.
44:4, 343–350
(2000).
64. G. J. Braun and M. D. Fairchild, Gamut Mapping for Pictorial Im-
ages, in
TAGA Proc.
, TAGA, Rochester, NY, 645–660 (1999)
65. B. H. Kang, M. S. Cho, J. Morovic, and M. R. Luo, Gamut Compres-
sion Analysis Based on Observer Experimental Data, in
Proc. 7
th
IS&T/SID Color Imaging Conf.
, IS&T, Springfield, VA, 295–300 (1999)
66. B. H. Kang, M. S. Cho, J. Morovic, and M. R. Luo, Gamut Compres-
sion Algorithm Development on the Basis of Observer Experimental
Data, in
Proc. 8
th
IS&T/SID Color Imaging Conf.
, IS&T, Springfield,
VA, 268–272 (1999)
67. CARISMA,
Color Appearance Research for Interactive System Man-
agement and Application—CARISMA
, Work Package 2—Device
Characterization, Report WP2–19 Color Gamut Compression (1992)
68. K. Schläpfer,
Color Gamut Compression—Correlations Between Cal-
culated and Measured Values
, IFRA Project, EMPA, Switzerland, 8
August 1994
69. H. Motomura, O. Yamada and T. Fumoto, Categorical Color Map-
ping for Gamut Mapping, in
Proc. 5
th
IS&T/SID Color Imaging Conf.
,
IS&T, Springfield, VA, 50–55 (1997)
70. H. Motomura, Categorical Color Mapping for Gamut Mapping II—
Using Block Average Image, in
SPIE Proc.
, SPIE, Bellingham, WA,
3648, 108—119 (1999)
71. H. Motomura, Gamut Mapping using Color-Categorical Weighting
Method, in
Proc. 8
th
IS&T/SID Color Imaging Conf.
, IS&T, Spring-
field, VA,318–323 (2000)
72. H. Haneishi, K. Miyata, H. Yaguchi, and Y. Miyake, A New Method
for Color Correction in Hardcopy from CRT Images,
J. Imaging
Technol.
37:1, 30–36 (1993)
73. M. Wolski, J. P. Allebach and C. A. Bouman, Gamut Mapping.
Squeezing the Most out of Your Color System, in
Proc. 2
nd
IS&T/SID
Color Imaging Conf.
, IS&T, Springfield, VA, 89–92 (1994)
74. E. M. Granger, Gamut Mapping for Hard Copy Using the ATD Color
Space, in
SPIE Proc.
, SPIE, Bellingham, WA, 2414, 27–35 (1995)
75. K. E. Spaulding, R. N. Ellson and J. R. Sullivan, UltraColor: A New
Gamut Mapping Strategy, in
SPIE Proc.
, SPIE, Bellingham, WA,
2414, 61–68 (1995)
76. S. D. Kim, C. H. Lee, K. M. Kim, C. S. Lee, and Y. H. Ha, Image
Dependent Gamut Mapping Using a Variable Anchor Point, in
SPIE
Proc.
, SPIE, Bellingham, WA, 3300, 129–137 (1998)
290 Journal of Imaging Science and Technology® Morovic and Luo
77. C. S. Lee, C. H. Lee and Y. H. Ha, Parametric Gamut Mapping Algo-
rithms Using Variable Anchor Points,
J. Imaging Sci. Technol.
44:1,
68–73 (2000)
78. N. Katoh and M. Ito, Applying Non-linear Compression to the Three-
dimensional Gamut Mapping,
J. Imaging Sci. Technol.
44:4, 328–333
(2000)
79. N. Katoh and M. Ito, Applying Non-linear Compression to the Three-
Dimensional Gamut Mapping, in
Proc. 7
th
IS&T/SID Color Imaging Conf.
,
IS&T, Springfield, VA, 155–159 (1999)
80. N. Katoh, M. Ito and S. Ohno, Three-dimensional gamut mapping us-
ing various color difference formulae and color spaces,
J. Electronic
Imaging
, 8:4, 365-379 (1999)
81. P. G. Herzog and H. Büring, Optimizing Gamut Mapping: Lightness
and Hue Adjustments,
J. Imaging Sci. Technol.
44:4, 334–342 (2000)
82. P. G. Herzog and H. Büring, Optimizing Gamut Mapping: Lightness
and Hue Adjustments, in
Proc. 7
th
IS&T/SID Color Imaging Conf.
, IS&T,
Springfield, VA, 160–166 (1999)
83. L. W. M. MacDonald, J. Morovic and K. Xiao, A Topographical Gamut
Mapping Algorithm Based on Experimental Observer Data, in
Proc.
8
h
IS&T/SID Color Imaging Conf.
, IS&T, Springfield, VA, 311–317
(2000)
84. M. Ito and N. Katoh, Three-dimensional Gamut Mapping Using Vari-
ous Color Difference Formulæ and Color Spaces, in
SPIE Proc.
, SPIE,
Bellingham, WA, 3648, 83–85 (1999)
85. CIE,
Industrial Color-Difference Evaluation
, CIE 116–1995 (1995)
86. M. R. Luo and B. Rigg, BFD(l:c) Color Difference Formula, Part 1 and
Part 2,
J. Soc. Dyers Colourists
126–132 (1987)
87. R. Balasubramanian, R. deQueiroz, R. Eschbach, and W. Wu, Gamut
Mapping to Preserve Spatial Luminance Variations, in
Proc. 8
th
IS&T/
SID Color Imaging Conf.
, IS&T, Springfield, VA, 122–128 (2000)
88. K. M. Braun, R. Balasubramanian and S. J. Harrington, Gamut Map-
ping Techniques for Business Graphics, in
Proc. 7
th
IS&T/SID Color
Imaging Conf.
, IS&T, Springfield, VA, 149–154 (1999)
89. L. W. MacDonald, J. M. Deane and D. N. Rughani, Extending the
Color Gamut of Printed Images, in
Proc. of AIC/RPS Symposium
‘Images in Color’
, Cambridge, UK (1994)
90. P. Laihanen, Exact Soft Proofing,
J. Imaging Sci. Technol.
38:5, 432–
440 (1994)
91. F. K. Musgrave, Prisms and Rainbows: A Dispersion Model for Com-
puter Graphics,
Graphics Interface ’89
, 227–234 (1989)
92. B. Pham and G. Pringle, Color Correction for an Image Sequence,
IEEE Comp. Graph. Appl.
38–42 (1995)
