Content uploaded by Antonino Cosentino
Author content
All content in this area was uploaded by Antonino Cosentino on Apr 11, 2014
Content may be subject to copyright.
R E S E A R C H A R T I C L E Open Access
Identification of pigments by multispectral
imaging; a flowchart method
Antonino Cosentino
Abstract
The literature on the application of Multispectral and Hyperspectral imaging for identification of pigments on
artworks is sparse. While these methods do not provide the analytical capability that spectroscopies do offer, the
use of spectral imaging has the advantage of being a rapid and relatively low-cost solution for the examination of
large areas. This paper presents a flowchart for the identification of historical pigments applied with gum Arabic
using multispectral imaging (wavelength ranging from 360 to 1700 nm) performed with a modified digital camera
for infrared, visible and ultraviolet photography; and an InGaAs camera for infrared reflectography. The flowchart
method will be most successful on paint made of one layer of pure pigment, and it can selectively discriminate
only a fraction of the 56 pigments analyzed. Though, considerably limited in its analytical capabilities, the low cost
and speed of the workflow make the method worthwhile, even if only to localize retouching and areas appearing
the same hue but painted with different pigments. The InGaAs camera is the only expensive instrument used in this
study but its cost is relatively affordable for the average painting conservation studio since only a model with a low
pixel count is required (320×256 pixels) rather than a more sophisticated InGaAs scanner system.
Introduction
Multispectral imaging (MSI) [1,2] and Hyperspectral
Imaging [3-6], have been suggested as methods for the
non-destructive identification of pigments. Though, it is
mandatory to point out that these methods are problem-
atic and the user may be subjected to draw conclusions
that remain uncertain, essentially, because pigments are
often mixed and overlapped in layers to make the de-
sired color and effect.
To identify pigments with an acceptable degree of cer-
tainty, at least one other material specific technique
must be used to complement hyper or multispectral im-
aging diagnostics. The use of MSI to tentatively identify
pigments has an important advantage justifying its appli-
cation: the rapid and low-cost survey of large areas. The
intention of this paper is to show that with a flowchart
based methodology it is possible to tentatively identify
some historical pigments by means of MSI performed
with simplified equipment and without the aid of im-
aging analysis software. This method doesn’t claim to
allow the identification of all different pigments, but it
will work for those which present peculiar behaviors in
the range of the electromagnetic spectrum readily ob-
servable with an IR-VIS–UV modified digital camera
(360–1100 nm) and an InGaAs camera (900–1700 nm).
In this way, selected pigments are likely to be identified
by means of MSI examination. This simplified approach,
though demonstrated to be limited in its analytical diag-
nostic capabilities, has the benefit of being accessible
and easy to implement by professionals in the art con-
servation and examination field.
This method is more likely to succeed when applied on
artworks where pigments have been applied in one single
layer and not mixed; as is the case with miniatures [6,7],
drawings [8] and prints. Unlike other references, which sug-
gest the use of software algorithms to analyze the MSI im-
ages, this paper proposes a more straightforward method
simply based on visual examination and the use of a photo-
editing software for the characterization of features appar-
ent in the image.
Multispectral imaging
Imaging methods
This paper illustrates a flowchart method for pigment
identification based on the acquisition of MSI images in
4 spectral bands: Ultraviolet, UV (360–400 nm); Visible,
Correspondence: antoninocose@gmail.com
Independent Scholar at “Cultural Heritage Science Open Source”Blog,
chsopensource.org, Piazza Cantarella 11, Aci Sant’Antonio 95025, Italy
© 2014 Cosentino; licensee Chemistry Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly credited.
Cosentino Heritage Science 2014, 2:8
http://www.heritagesciencejournal.com/content/2/1/8
VIS (400–780 nm); Infrared, IR (780–1100 nm) and In-
frared Reflectography, IRR (1000–1700 nm), Figure 1.
The acronyms for the MSI methods presented in this
paper highlight first the spectral band followed by R
(Reflected), F (Fluorescence), FC (False Color). So the 8
imaging methods are called VIS (Visible), IR (Infrared),
UVF (UV Fluorescence), UVF254 (UVC light source),
UVR (UV Reflected), IRFC (Infrared False Color),
IRF (IR Fluorescence), IRR (Infrared Reflectography),
Figures 1 and 2.
There are a number of studies on the application of each
of the above mentioned imaging methods specifically for
the identification of pigments: UV Fluorescence (UVF)
[9-12], UV Reflected (UVR) [13], Infrared False Color
(IRFC) [14,15], Infrared Fluorescence [16-18] and Infrared
Reflectography [19]. Though, there is no comparative study
carried out using all of those methods, and therefore this
paper intends to fill that void.
Instrumentation
Imaging devices
The MSI images presented in this paper were acquired
with a Nikon D800 DSLR (36 MP, CMOS sensor) digital
camera modified for “full spectrum”, ultraviolet–visible‐
infrared photography (between about 360 and 1100 nm).
The CMOS sensor responds both to the near infrared
and near ultraviolet ranges of the spectrum, however
manufacturers install an IR cut-off filter in front of the
sensor to reduce infrared transmission. There are com-
panies that will remove this filter in commercial cameras
for a small fee, and then the camera is said to be “full
spectrum”. The camera is tethered to a computer to
allow sharp focusing in non‐visible modes (IR and UV)
using live view mode. The MSI images created with this
camera do not precisely image the same area in each
waveband because it is necessary to refocus the lens at
the different wavelengths. In order to allow a compara-
tive examination through the different spectral ranges,
those images are uploaded as layers of a single docu-
ment file in an image editing software, such as Adobe
Photoshop or GIMP, and manually resized to overlap
one another. Infrared Reflectography (IRR) was per-
formed with an InGaAs camera (320×256 pixels) Merlin
NIR by Indigo Systems. Some suggested references on
art documentation and examination using each spectral
band and the relative instrumentation for Infrared
[1,2,20-23] and Ultraviolet [24,25] are given.
Filters
The filters chosen for use in this paper are commercially
distributed for photography and so they are easy to find.
Their transmission spectra are available on the manufac-
turers’websites, Schneider Optics for the B + W filters,
Maxmax.com for the X-Nite filter and Heliopan for the
infrared filter. This is the filter set used for the MSI: a)
For Ultraviolet Reflected (UVR) photography, the B + W
403 filter is used together with the X-NiteCC1. B + W
403 allows just the UV light to pass, and X-NiteCC1 is
necessary to stop the IR produced from the UV lamp; b)
For Visible (VIS) photography, just the X-NiteCC1 filter
is sufficient; c) For UV Fluorescence (UVF) photography,
the B + W 420 must be mounted to stop the reflected
UV, and the X-NiteCC1 is also necessary to exclude any
infrared from the UV lamp; d) For Infrared (IR), Infrared
Fluorescence (IRF) and Infrared Reflectography, just the
Heliopan RG1000 is used.
Lenses
A Nikon Nikkor 50 mm f/1.8D AF lens was used for all
the MSI photos on the pigment swatches. Standard pho-
tography lenses are generally fine for MSI. Indeed, they
are transparent from 350 nm to 1700 nm. Though, it is
Figure 1 Illustration of the spectral bands, imaging methods and imaging devices which contribute to the flowchart described in
this paper.
Cosentino Heritage Science 2014, 2:8 Page 2 of 12
http://www.heritagesciencejournal.com/content/2/1/8
recommended to test the lenses for hot spots in the in-
frared imaging. Hot spots can be caused by a number of
factors, such as coatings and/or the interaction between
elements within the lens and the image sensor. Lists of
other lenses tested for hot spots are available online
[26]. It is also recommended to use fixed focal lenses
and avoid complex lenses that are likely to give flares in
the infrared and ultraviolet photography. Also it is rec-
ommended to avoid telephoto lenses over 200 mm for
MSI photography, so that the lens will be fast enough to
work with the low intensity emission produced in the
UV Fluorescence and IR Fluorescence imaging.
Lighting
Halogen lamps are used for VIS and IR photography.
For UV photography (UVF and UVR) high-Flux 365 nm
LED lamps are recommended. The advantages of UV
LED lighting are obvious: instant start up, no heating
up, very lightweight and sturdy. These LEDs are filtered
to cut off any visible light with a UV-pass glass analo-
gous to the B + W 403. This paper considers the UV
Fluorescence excited by a 254 nm UV lamp (UVF254) as
an additional and separate imaging method. These lamps
must be used with extreme care since this UV range
(UVC) is dangerous for the eyes and skin. Infrared fluor-
escence (IRF) requires illumination in the visible range
only, which can be implemented with a white light LED
lamp filtered with the X-Nite CC1 filter to cut off emis-
sion in the near-IR.
Calibration
For the camera settings and image processing of visible
images it is recommended to follow normal color man-
agement procedures [2]. For the purpose of this work
the American Institute of Conservation Photo Docu-
mentation (AIC PhD) target [27] was used for calibra-
tion of visible photography. The camera has been
calibrated with the X-rite ColorChecker Passport and its
bundled software. Due to differences in technologies and
variables in manufacturing processes every camera cap-
tures colors a bit differently. Even two identical cameras
from the same company are a bit different. Even if the
Raw processing software could have included a profile
for the specific model camera, it is possible to get even
better results by creating a profile specifically for the
RAW output of the particular camera. ColorChecker
Passport Classic Target is the industry standard color
reference target for creating DNG profiles and for evalu-
ating specific colors. It is important to point out that the
camera calibration is another step, in addition to the
common white balance performed even on compact
cameras. The first one corrects for different rendering of
colors by different cameras, while the second one cor-
rects for different lighting. The images were shot RAW
and they were then color corrected using the camera
profile above mentioned and white balanced using the
N6.5 neutral grey patch in the AIC target. The grey
patches are identified by the following designations
(white to black): white; N8; N6.5; N5; N3.5; and black.
They were also exposure corrected: N8 patch 150 +/−5
Figure 2 Madonna and Child, Ingels Collection, Sweden. Example of Multispectral Imaging documentation and the acronyms used in
this paper.
Cosentino Heritage Science 2014, 2:8 Page 3 of 12
http://www.heritagesciencejournal.com/content/2/1/8
for VIS. The same patch is also used for correcting the
other images: 100 +/−5 for IR and IRR, and 50 for
UVR. On the other hand, calibration for ultraviolet
fluorescence is trickier as there is not yet an officially
recognized reference standard for UVF and UVF254
photography. As an aid for color balancing and in order
to guarantee reproducibility of the UVF photos, the AIC
PhD target was accompanied with 3 emitters activated
by the LED UV and UV254 lamps: a section of a card
for forensic UV photography (orange fluorescence), a
swatch of zinc white (yellow fluorescence), and a fluores-
cent paint (green fluorescence). These 3 UV emitters to-
gether with the red fluorescence emission of the red
square of the AIC PhD target itself are used for color
balancing of UVF and UVF254 photos. This is per-
formed on the RAW files editing them with Adobe
Camera RaW 7.1 and assigning Temperature 15000
Kelvin and Tint 70. The images are also exposure cor-
rected using the red fluorescence of the red patch: Red
channel 70 +/−5, Green 0, Blue 0. In order to check the
correct exposure for IRF photography a swatch of cad-
mium red was also added to the UV emitters, and the
IRF images are exposure corrected with the red cad-
mium swatch at RGB 30. Figure 3 shows the 8 MSI
images of the AIC target together with the UV and IR
fluorescence emitters. The Infrared False Color image is
made by digitally editing the VIS and IR images. A copy
of the VIS image is edited to become the IRFC image.
The VIS green channel substitutes the blue channel and
the red channel the green channel. Then, the IR image
constitutes the red channel of the edited VIS.
Pigments
A collection of swatches of 56 historical pigments (Figure 4)
have been applied using gum Arabic as a binder on a cellu-
lose and cotton watercolor paper, acids and lignin free,
commercialized by “Fabriano”, 270 gr/m
2
. This paper is not
treated with optical brighteners, it’s slightly UV Fluorescent,
and it reflects IR. Two cross-hair lines, 0,2 mm (vertical)
and 0.4 mm (horizontal), were printed on each swatch of
paper before the application of paint, in order to have a
means to evaluate the pigment transparency in the IR and
IRRimaging.GumArabicwaschosenasabinderbecause
it is the medium used in watercolor technique where pig-
ments are more often used pure. Therefore, these are the
artworks where the proposed method is likely to deliver
correct identification, rather than in other painting tech-
nique, as linseed oil, were pigments are more expected to
be blended and layered. The pigments were mulled into the
binder which was added as needed for each pigment and
applied with brush. No other means to control and meas-
ure thickness of the paint and ratio binder-pigment was im-
plemented since this study wants to check the capabilities
of this flowchart method on actual artworks where all those
parameters are not likely to be easily determined. Some pig-
ments that are not actually used with gum Arabic were in-
cluded in this paper since it is planned to use the same set
of pigments for future studies with other binders. Table 1
Figure 3 The AIC PhD target (left) was accessorized with swatches of UV and IR fluorescent paints (right) to aid calibration of UVF and
IRF photography. Both targets are necessary for check the correct shooting and post-processing of all the 8 imaging methods.
Cosentino Heritage Science 2014, 2:8 Page 4 of 12
http://www.heritagesciencejournal.com/content/2/1/8
shows the list of pigments used in this study. Among all the
pigments and their varieties ever used in art these 56-
pigments collection is not exhaustive but it attempts, at
least, to be a selection of the most used ones from antiquity
to early 1950’. A swatch of just gum Arabic is added as a
reference for the binder.
All the pigments are distributed by Kremer Pigments and
their relative information about composition and manufac-
turing can be found on the Kremer website [28] searching
for the specified product code. The high resolution multi-
spectral images of the pigment swatches can be accessed
online [29] and browsed on an IIPImage [30] server system
Figure 4 Multispectral images of 56 historical pigments laid with gum Arabic on watercolor paper.
Table 1 List of pigments with the Kremer Pigments product code
BLACKS Cobalt violet, 45800 REDS Chalk, 58000
Ivory black, 12000 BROWNS Alizarin, 23600 YELLOWS
Vine black, 47000 Burnt Sienna, 40430 Cadmium red, 21120 Cadmium yellow, 21010
Bone black, 47100 Burnt umber, 47010 Red lead, 42500 Cobalt yellow, 43500
Lamp black, 47250 Van Dyke brown 41000 Read ochre, 11574 Lead Tin yellow I, 10100
BLUES Raw Sienna, 17050 Vermilion, 10610 Lead Tin yellow II, 10120
Azurite, 10200 Raw umber, 40610 Madder lake, 372051 Massicot, 43010
Blue bice, 10184 GREENS Lac dye, 36020 Naples yellow, 10130
Cobalt blue, 45730 Cadmium green, 44510 Carmine lake, 42100 Orpiment, 10700
Egyptian blue, 10060 Chrome green, 44200 Realgar, 10800 Saffron, 36300
Indigo, 36007 Cobalt green, 44200 WHITES Yellow ochre, 40010
Maya blue, 36007 Green earth, 11000 Lead white, 46000 Yellow Lake reseda, 36262
Prussian blue, 45202 Malachite, 10300 Zinc white, 46300 Gamboge, 37050
Smalt, 10000 Phthalo green, 23000 Lithopone, 46100
Ultramarine nat, 10510 Verdigris, 44450 Titanium white, 46200
Phthalo blue, 23050 Viridian, 44250 Gypsum, 58300
Cosentino Heritage Science 2014, 2:8 Page 5 of 12
http://www.heritagesciencejournal.com/content/2/1/8
that allows the user to blend between the multispectral
images for immediate comparison.
Flowchart
1st step. VIS
The logic of this flowchart is to start with the most
straightforward imaging methods and move toward the
less specific ones. Rectangular shapes list the pigments
showing the same feature such as, for example, a red
UV fluorescence. If the same group of pigments shows
also other properties in common, these are listed in the
bottom of the same rectangular shape. The first step is
to assign the pigment to one of these color categories:
white, black, blue, green, yellow, brown, red. The flow-
chart is likely to succeed for the identification of paints
made of just one pure pigment. If the paint is a mixture
or a glaze, multispectral imaging loses its identification
capacity. In that case, it’s recommended to first perform
an examination under magnification, with at least a
hand-held USB microscope, to check if the paint is a
mixture. Particular attention should be directed towards
greens and violets. Indeed, often they are a mix of blue
and yellow for the green, and red and blue for the violet.
They could also be a layered glaze, yellow over blue for
the greens and red over blue for the violets.
2nd step. UVR
UV light interacts just with the very surface layer of
paint, so the UVR image is specific to the topmost pig-
ment and layers of paint underneath do not influence it.
Furthermore, a uniform varnish coating doesn’t interfere
with the pigments’appearance in the UVR but it will
absorbs some UV causing just a change in the overall
brightness. Strong differences in the brightness of UV
reflected images such as that of lithopone and titanium
white are still observable, Figure 5. Two categories are
defined: bright, and dark. A pigment is defined as UVR
bright if it reflects more UV than the AIC target N8
patch, i.e. its RGB > 60.
3rd step. UVF
As described for the UVR imaging, UV fluorescence
comes just from the surface of the paint layer and it isn’t
influenced by layers underneath. Though in this case,
the varnish does play a major role since it generally
exhibits strong fluorescence and could overwhelm the
actual fluorescence of the pigments. Consequently, MSI
Figure 5 Lithopone and titanium white swatches respectively reflect and absorb UV light. The dammar varnish coating doesn’t affect the
UV reflected image of the two pigments.
Figure 6 Prussian blue is IR absorbent. In the infrared image phthalo blue appears bright because it is IR reflective. On the other hand, Indigo
appears bright because it is IR transparent. Indeed, when it is overlapped with Prussian blue it looks dark. Consequently, the infrared false color of
the indigo swatch looks not the ordinary bright red but as dark as Prussian blue.
Cosentino Heritage Science 2014, 2:8 Page 6 of 12
http://www.heritagesciencejournal.com/content/2/1/8
documentation is recommended when the varnish is
removed from the artwork. These decision categories are
defined for the UV fluorescence: none (meaning no
sizeable fluorescence is observed, i.e. RGB < 30), white,
blue, red, orange, yellow.
4th step. UVF 254
A 254 nm UV lamp allows the observation of fluorescence
otherwise not excited by the 365 nm UV lamp discussed in
the 3rd step, such as that which particularly occurs with
madder lake. The same decision steps as for UVF are de-
fined: none (meaning no sizeable fluorescence is observed),
red, orange, yellow, white. Again, as for the UVF, these
color categories are to be intended liberally since the
perception of the color is not well defined.
5th step. IRF
Only a few pigments exhibit Infrared Fluorescence, namely
cadmium based pigments and Egyptian blue, making it a
highly specific method. Just two decision steps: yes (RGB >
10), or none.
6th step. IR
We can classify the behavior of each pigment in respect
to the IR region as transparent, reflective or absorbent.
When observing the IR image of the pigment swatches,
if the pigment is IR transparent we can distinguish the
carbon ink line, if it is IR reflective, the pigment will
look bright and the ink line will be not visible. The pig-
ment could also be IR absorbent and in this case the ink
could not be detected since the swatch will appear pitch
black. We can determine the IR behavior of a pigment
with reasonable certainty only if it is applied on an IR
reflective ground. Otherwise the under-paint will play a
major role in the IR appearance of the above layer of
paint. Note the examples in Figure 6, Prussian blue ab-
sorbs all the infrared so it is pitch dark in IR imaging.
Phthalo blue and indigo both appear bright. Phthalo blue
Figure 7 Cobalt-based pigments, such as cobalt blue, show an inverted behavior in the IRR. Differently than the other pigments, they
become more “dark”(absorbing) in the IRR, with respect to the IR imaging.
Figure 8 Flowchart for the identification of white pigments by multispectral imaging.
Cosentino Heritage Science 2014, 2:8 Page 7 of 12
http://www.heritagesciencejournal.com/content/2/1/8
reflects IR, while indigo is transparent to the IR and
therefore the white ground appears beneath. So, while
phthalo blue would look bright on any ground, indigo
would be bright or dark depending on the under paint.
So, since we cannot know in advance if the pigment is
infrared transparent, there is no way to know if its
brightness or darkness in the IR image is due to the
under layer. With the infrared imaging we are indeed on
more uncertain ground when it comes to the identifica-
tion of layered paint. Consequently, two simple categor-
ies are defined for the IR imaging: bright, dark. In order
to discriminate among more pigments it is necessary to
define two specific IR bright threshold values, one for
the yellows (RGB > 140) and one for all the other colors
(RGB > 100).
7th step. IRR
At longer IR wavelengths, such as those captured with an
InGaAs camera for infrared Reflectography (IRR) most of
thepigmentsbecomeabitmoretransparentthanintheIR
imaging. Cobalt-based pigments exhibit the property most
useful for pigment identification as they become less trans-
parent in the IRR, Figure 7. Looking at the change of
brightness in respect to the IR imaging, two categories are
defined: darker (Delta RGB > 50), and same (which encom-
passes the normal added transparency in the IRR range).
Figure 9 Flowchart for the identification of blue pigments by multispectral imaging.
Figure 10 Flowchart for the identification of green pigments by multispectral imaging.
Cosentino Heritage Science 2014, 2:8 Page 8 of 12
http://www.heritagesciencejournal.com/content/2/1/8
8th step. IRFC
For the last step comes Infrared false color. As we saw in
Figure 6, the infrared false color appearance is strongly
influenced by the under layer, and as for the IR step this
method is useful only for single layer paint examination.
The categories are defined for the color produced as: red,
pink,green,black,white,andpurple.
Whites
Four white pigments have been tested and also two
white minerals, gypsum and chalk, which are not prop-
erly designated as pigments because of their low hiding
power, but are known to be used in art. These last ones
turn grey when mixed with gum Arabic and require
many applications to acquire enough opacity. Of the
white pigments, only titanium white and zinc white
show UV absorbance, and the latter can be identified for
its yellow UV Fluorescence, Figure 8.
Blacks
The four most used carbon-based blacks were tested and,
as expected, none of the imaging methods can distinguish
among them.
Blues
Eleven blue pigments, including Cobalt violet, were tested.
It is notable that cobalt containing pigments reflect UV
(cobalt violet, cobalt blue, smalt). Also, smalt and cobalt
blue turn dark in the IRR imaging, spectra are available
[31], Figure 9. Natural ultramarine exhibits a characteristic
white UV254 fluorescence.
Figure 11 Flowchart for the identification of yellow pigments by multispectral imaging.
Figure 12 Flowchart for the identification of red pigments by multispectral imaging.
Cosentino Heritage Science 2014, 2:8 Page 9 of 12
http://www.heritagesciencejournal.com/content/2/1/8
Greens
Eight green pigments were tested. Cadmium green is easily
identified for its IR Fluorescence and cobalt green due to
its darker IRR image, Figure 10.
Yellow
Eleven yellow pigments were tested. UV Fluorescence is
an intense red for Cadmium yellow, while it is yellow for
Saffron and the yellow lake made of reseda. Gamboge
exhibits an intense red with UV254, Figure 11.
Reds
Nine red pigments were tested, Figure 12. Cadmium red
is identified for its exceptionally bright IRF, however for
mineral pigments the composition is likely to be rela-
tively consistent, and with organic lakes there could be
variation in the composition; such as the case with the
nine madder lakes varieties produced by Kremer (prod-
uct codes 37202, 37203, 37217, 37218, 372051, 372052,
372057, 372058, 372142). They were all tested and
372051 was chosen for this flowchart since the fluores-
cence exhibited under UV254 illumination was the most
representative when compared with the madder lakes
previously documented in literature [32-35], Figure 13.
Consequently, while the presence of the UVF254 fluor-
escence is a positive test for madder lake, its absence
doesn’t rule this lake out.
Browns
Five brown pigments were tested. As expected, none of
the imaging methods can distinguish among them.
A case study
The Ingels Collection, Sweden, owns a Madonna and
Child (painting on panel), Figure 2, which is currently
being studied and has been documented with multispec-
tral imaging by the author. The painting seems to date
back to the early 14th century Sienese school, as pro-
posed by some scholars. The lips of the Madonna are
interesting to test the flowchart, see Figure 14. The lips
Figure 13 Kremer Pigments produces 9 varieties of madder lake. Only one product (372051) shows the characteristic orange UVF254 Fluorescence.
Figure 14 The flowchart method helps to correlate the 7 multispectral imaging photos of the lips of the Madonna and Child panel,
Ingels Collection, Sweden in attempt to arrive at pigment identification in the case of a mixture.
Cosentino Heritage Science 2014, 2:8 Page 10 of 12
http://www.heritagesciencejournal.com/content/2/1/8
are retouched with a red pigment that reflects UV, has a
pink UV fluorescence, is IR and IRR transparent, and IR
fluorescent. Not a single pigment from the selection of
reds tested here shows all these features, in particular
the UVF together with the IRF. So, it seems the lips have
been retouched with a mix of cadmium red, which
accounts for the IRF, and madder lake for the pinkish
UVF. This hypothesis seems to be supported by the mi-
cro image of an area on the lips where different red pig-
ments are observed. Even if the pigment identification
cannot be accomplished in the majority of the cases,
MSI is extremely valuable to differentiate pigments of the
same hue, which have been used in subsequent retouch-
ings of an area. The mantle of the Madonna was restored
at least two times using two different pigments, Figure 15.
The original blue and the two blue pigments can be local-
ized because they behave differently in the MSI images.
Retouching 2 is characterized by its intense UVF which is
due to its binder, while retouching 1 is differentiated from
the original blue (likely azurite) because is bright in the IR.
Conclusions
This paper represents the first attempt to correlate MSI
methods to achieve a preliminary pigment identification.
This study focused on pigments applied with gum
Arabic, which is the medium used most in watercolor, a
technique where pigments are less likely to be blended
and overlapped. Further work is necessary to take into
account the effects of other mediums, linseed oil, tem-
pera and acrylic, and degree of aging.
It was shown that some particularly useful features
such as UV Fluorescence, IR Fluorescence and the
losses of infrared reflectance in the IRR imaging allow
for rather reliable identification of certain pigments.
On the other hand this MSI flowchart should be seen
as complementary to analytical methods, in particular,
elemental spectroscopies, such as X-ray Fluorescence
spectroscopy. For example, on the blue pigments
flowchart it is showed that Prussian blue and azurite
are just distinguishable by their IRFC, which is the
less reliable discriminant. Though, XRF can easily dis-
tinguish among them through the presence or ab-
sence of significant copper content.
Unfortunately, it is not always possible to perform all the
8 imaging methods to study an artwork. This is the case of
paintings with the old varnish whose UVF can overwhelm
that of the pigments, and the case where it is not possible
to have total darkness to correctly execute IRF and UVF. In
these cases, the flowchart can be still followed bypassing
the missing steps, but losing, of course, part of its identifica-
tion capability.
It would be valuable to compare this flowchart approach
with other spectral imaging technologies [36,37], methods
which are currently investigated and which use estimation
techniques for visible spectrum imaging [38], and could be-
come more powerful with the implementation of liquid-
crystal tunable filters or interference filters. It would be also
worth to apply the flowchart approach to other conserva-
tion studies on which Hyperspectral Imaging has proven
successful such as on inks [39,40] and parchment [41].
Competing interests
The author declares that he has no competing interests.
Acknowledgements
This work has been possible thanks to Kremer Pigments who has provided
the pigment samples. Thanks also to the Ingels Collection, Sweden for
permitting the use of the material on the Madonna and Child panel.
Figure 15 MSI is useful to localize retouchings with two different blue pigments on the Madonna and Child panel, Ingels Collection, Sweden.
Cosentino Heritage Science 2014, 2:8 Page 11 of 12
http://www.heritagesciencejournal.com/content/2/1/8
Received: 30 July 2013 Accepted: 6 March 2014
Published: 17 March 2014
References
1. Cosentino A: A practical guide to panoramic multispectral imaging.
e-Conservation Magazine 2013, 25:64–73.
2. Frey FS, Warda J, American Institute for Conservation of Historic and Artistic
Works, Digital Photographic Documentation Task Force: The AIC guide to
digital photography and conservation documentation. Washington, D.C:
American Institute for Conservation of Historic and Artistic Works; 2008.
3. Zhao Y, Berns RS, Taplin LA, Coddington J: An investigation of
multispectral imaging for the mapping of pigments in paintings. In Proc.
SPIE 6810. San Jose, CA: Computer Image Analysis in the Study of Art; 2008.
4. Kubik M: Chapter 5 hyperspectral imaging: a new technique for the
non-invasive study of artworks. In Physical Techniques in the Study of Art,
Archaeology and Cultural Heritage, volume 2. Oxford, UK: Elsevier;
2007:199–259.
5. Padoan R, Steemers T, Klein M, Aalderink B, de Bruin G: Quantitative
hyperspectral imaging of historic documents. In 9th International
Conference on NDT of Art. Jerusalem, Israel; 2008.
6. Isacco E, Darrah J: The ultraviolet-infrared method of analysis, a scientific
approach to the study of Indian miniatures. Artibus Asiae 1993,
53(3/4):470–491.
7. Melessanaki K, Papadakis V, Balas C, Anglos D: Laser induced breakdown
spectroscopy and hyper-spectral imaging analysis of pigments on an
illuminated manuscript. Spectrochim Acta B At Spectrosc 2001, 56:2337–2346.
8. Havermans J, Aziz HA, Scholten H: Non destructive detection of iron-gall
inks by means of multispectral imaging. Restaurator 2003, 24:88–94.
9. Buzit Tragni C: The use of ultraviolet-induced visible fluorescence for examin-
ation of photographs, Advanced residency program in photograph
conservation, Image Permanence Institute, Rochester Institute of
Technology; 2005. http://notesonphotographs.org/images/d/df/
Claire_Tragni_for_web.pdf.
10. René de la Rie E: Fluorescence of paint and varnish layers (part I).
Stud Conserv 1982, 27(1):1–7.
11. René de la Rie E: Fluorescence of paint and varnish layers (part III).
Stud Conserv 1982, 27(3):102–108.
12. Comelli D, Valentini G, Nevin A, Farina A, Toniolo L, Cubeddu R: A portable
UV-fluorescence multispectral imaging system for the analysis of painted
surfaces. Rev Sci Instrum 2008, 79:086112.
13. Aldrovandi A, Buzzegoli E, Keller A, Kunzelman D: Investigation of painted
surfaces with a reflected UV false color technique. Italy: art’05, 8th
International Conference on Non Destructive Investigations and
Micronalysis for the Diagnostics and Conservation of the Cultural and
Environmental Heritage Lecce; 2005.
14. Moon T, Schilling MR, Thirkettle S: A note on the use of false-color infrared
photography in conservation. Stud Conserv 1992, 37(1):42–52.
15. Hoeniger C: The identification of blue pigments in early Sienese
paintings by color infrared photography. J Am Inst Conserv 1991,
30(2):115–124.
16. Bridgman CF, Gibson HL: Infrared luminescence in the photographic
examination of paintings and other art objects. Stud Conserv 1963,
8(3):77–83.
17. Accorsi G, Verri G, Bolognesi M, Armaroli N, Clementi C, Miliani C, Romani A:
The exceptional near-infrared luminescence properties of cuprorivaite
(Egyptian blue). Chem Commun 2009, 23:3392–3394.
18. Thoury M, Delaney JK, De la Rie ER, Palmer M, Morales K, Krueger J:
Near-infrared luminescence of cadmium pigments: in situ identification
and mapping in paintings. Appl Spectrosc 2011, 65(8):939–951.
19. Gargano M, Ludwig N, Poldi G: A new methodology for comparing IR
reflectographic systems. Infrared Phys Technol 2007, 49:249–253.
20. Mairinger F: The infrared examination of paintings. In Radiation in Art and
archeometry. Edited by Creagh DC, Bradley DA. Elsevier; 2000:40–55.
21. Mayer JW: Infrared reflectography and hiding thickness. In The Science of
Paintings. New York: Springer-Verlag; 2000:125–127.
22. van Asperen JRJ, Boer d: Reflectography of paintings using an infrared
vidicon television system. Stud Conserv 1969, 14(3):96–118.
23. Falco CM: High-resolution infrared imaging. San Diego: SPIE Optics+Photonics
Conference; 2010.
24. Savage G: Forgeries, fakes, and reproductions, a handbook for collectors,
Volume 3. London: White Lion Publishers Ltd; 1976.
25. Rorimer JJ: Ultraviolet rays and their use in the examination of works of art.
1st edition. New York: Metropolitan Museum of Art; 1931.
26. Hannemyr G, IR L: What lenses are suitable for IR photography. DPanswers.
com http://dpanswers.com/content/irphoto_lenses.php.
27. AIC PhotoDocumentation Targets (AIC PhD Targets), conservation‐us.org.
http://www.rmimaging.com/aic_phd.html.
28. Kremer Pigments Inc: Kremer Pigments Inc. http://kremerpigments.com/.
29. Cultural Heritage Science Open Source”.Blog. References/Pigments page
http://chsopensource.org/multispectral-imaging-pigments/.
30. IIPImage server system. http://iipimage.sourceforge.net/.
31. Lawrence Berkeley National Laboratory Pigment Database. http://coolcolors.
lbl.gov/LBNL-Pigment-Database/database.html.
32. Kirby J, Spring M, Higgit C: The technology of eighteenth and nineteenth
century Red lake pigments. National Gallery Tech Bull 2007, 28:69–95.
33. Kirby J, Spring M, Higgit C: The technology of Red lake pigment
manufacture: study of the dyestuff substrate. National Gallery Tech Bull
2005, 26:71–87.
34. FitzHugh EW: Artists’Pigments: A Handbook of Their History and
Characteristics. Washington, DC: National Gallery of Art; 1997.
35. Art Institute Chicago: UV Image: Madder lake pigment revealed. http://www.
artic.edu/aic/resources/resource/572.
36. Fischer C, Kakoulli I: Multispectral and hyperspectral imaging technologies in
conservation: current research and potential applications. Stud Conserv
2006, 14(Supplement 1):3–16(14).
37. Casini A, Lotti F, Baronti S, Castagna P, Stefani L: An image spectroscopy
system for the analysis and documentation of paintings. Sci Tech Cult
Herit 1992, 1:33–43.
38. Berns RS, Imai FH: The use of multi-channel visible spectrum imaging for pig-
ment identification. Rio De Janeiro Preprints: ICOM Committee for
Conservation, 13th triennial meeting; 2002.
39. Tse S, Goltz D, Guild S, Orlandini V, Trojan-bedynski M, Richardson M: Effect
of aqueous treatments on nineteenth-century iron-gall-Ink documents:
assessment using hyperspectral imaging. Book Paper Group Ann 2009,
28:75.
40. Havermans J, Aziz HA, Scholten H: Non destructive detection of iron-gall
inks by means of multispectral imaging part 2: application on original
objects affected with iron-gall-Ink corrosion. Restaurator 2008, 24(2):88–94.
41. Giacometti A, Campagnolo A, MacDonald L, Mahony S, Terras M, Robson S,
Weyrich T, Gibson A: Cultural Heritage Destruction: Documenting Parchment
Degradation via Multispectral Imaging. London: Electronic Visualisation and
the Arts (EVA); 2012.
doi:10.1186/2050-7445-2-8
Cite this article as: Cosentino: Identification of pigments by
multispectral imaging; a flowchart method. Heritage Science 2014 2:8.
Open access provides opportunities to our
colleagues in other parts of the globe, by allowing
anyone to view the content free of charge.
Publish with ChemistryCentral and every
scientist can read your work free of charge
W. Jeffery Hurst, The Hershey Company.
available free of charge to the entire scientific community
peer reviewed and published immediately upon acceptance
cited in PubMed and archived on PubMed Central
yours you keep the copyright
Submit your manuscript here:
http://www.chemistrycentral.com/manuscript/
Cosentino Heritage Science 2014, 2:8 Page 12 of 12
http://www.heritagesciencejournal.com/content/2/1/8
