Content uploaded by Giuseppe Capobianco
Author content
All content in this area was uploaded by Giuseppe Capobianco on Apr 01, 2014
Content may be subject to copyright.
VIII Congresso Nazionale di Archeometria
Scienze e Beni Culturali: stato dell’arte e prospettive
Bologna 5 - 7 Febbraio 2014
HYPERSPECTRAL IMAGING FOR EARLY DETECTION OF ALTERATION
PHENOMENA IN PAINT LAYERS
G. Capobianco 1*, F. Prestileo2, S. Serranti1, G. Bonifazi1
1 Department of Chemical Engineering Materials & Environment, Sapienza – University
of Rome, Italy, giuseppe.capobianco@uniroma1.it
2 CNR ICVBC - Institute for the Conservation and Valorization of Cultural Heritage,
Rome, Italy
Abstract
Hyperspectral imaging (HSI) is a technique that can be applied in various fields of science
and recent developments in hardware and software allow to increase its power and to decrease
the costs of the analysis based on reflectance spectrum in the visible and near infrared
wavelength range. This monitoring technique is particularly suitable for pigment
classification and for the monitoring of pictorial materials without the need of sampling [1].
In order to study the alteration of paint layers by solar aging , samples were prepared with
different pigments, binders and supports and artificially aged using solar box. The results
obtained for a sample composed by yellow ocher mixed with Arabic gum and applied to a
canvas support are presented in this work [2].The sample was acquired in the VIS-NIR (400-
1000 nm) and SWIR (1000-2500 nm) ranges by HSI after196 hours and 400 hours. Besides,
for comparison, color measurements were carried out before and after aging the sample to
monitor color changes during the two steps. The goal of this study is to verify the possibility
to detect alteration phenomena before they become macroscopically visible by HSI in the
VIS-NIR and SWIR ranges.
Materials and methods
The yellow ocher mixed with Arabic gum was applied on a canvas with a primer composed of
gypsum and calcium. After 30 days the following measures were carried out:
Colorimetric measurements by a Konica Minolta CM2600d spectrophotometer
with the following setup: CIE L * a * b * color scales, illuminant D65, 10 °
standard observer, geometry of measurement 45 °/0 °.
Hyperspectral imaging was carried out in two ways:
1) The first acquisition was carried out in the VIS -NIR range from 400 to 1000 nm
by means of an ImSpector model V10E (Specim, Finland), mounted on a
stereomicroscope.
2) another Specim system, the SISUCHEMA XL, was used, equipped with a
hyperspectral camera sensitive in the range from 1000 to 2500 nm with a
resolution sample / pixel of about 6 nm.
Chemometric analysis was performed in both cases using the PLS Toolbox (C) by Eigenvector
Research, Inc. Version: 7.3 Release: (16217) running on MATLAB Version 7.10.0.499
(R2010a) [3]. .
VIII Congresso Nazionale di Archeometria
Scienze e Beni Culturali: stato dell’arte e prospettive
Bologna 5 - 7 Febbraio 2014
Colorimetric measurements results
The PCA analysis of color measurements (Fig. 1) carried out using the spectrophotometer
shows a color change (indicating an alteration of the pictorial surface) with a thinning of the
measurement cloud points performed after 400 aging hours.
Fig. 1. Score plot of the colorimetric data processed by PCA.
VIS-NIR HSI results
The reflectance spectra (Fig. 2a) of the sample aged for 17 days shows slight reflectance
changes varying the irradiation time. Applying PCA (Fig. 2b) the different aging times of the
samples are highlighted as well as the detachment areas.
VIII Congresso Nazionale di Archeometria
Scienze e Beni Culturali: stato dell’arte e prospettive
Bologna 5 - 7 Febbraio 2014
(A)
(B)
Fig. 2. A) average spectra of the acquisitions performed on the sample in VIS-NIR range; B) score plot of PCA performed on
the sample at different aging times.
Selecting small ROIs (Region of Interest) on areas characterized by pigment and binder
detachments before aging, it was possible through PLS-DA classification to automatically
recognize all the alterations present in the investigated sample (Fig. 3).
VIII Congresso Nazionale di Archeometria
Scienze e Beni Culturali: stato dell’arte e prospettive
Bologna 5 - 7 Febbraio 2014
Fig. 3. Mapping carried out on the area analysed at different aging times by PLS-DA classification.
HSI SWIR results
The average spectra in the SWIR range of the samples aged at different times (Fig. 4a) show
the main differences in the region from 2200 nm, probably caused by a reduced thickness of
the paint layer and the increase of the underlying layer of preparation. The PCA analysis (Fig.
4b) highlights on the sample before aging some anomalies due to an incomplete adherence of
the pigment to the support.
(A)
VIII Congresso Nazionale di Archeometria
Scienze e Beni Culturali: stato dell’arte e prospettive
Bologna 5 - 7 Febbraio 2014
B)
Fig. 4. A) average spectra of the acquisitions performed on the sample in SWIR range; B) score plot of PCA performed on
the sample at different aging times
To check the distribution of the paint layer alteration varying the irradiation time, yellow
ochre powder has been acquired and processed and it has been used to set the model. Using
PLS-DA (Fig. 5) it was possible to verify the alteration of the pigment varying the irradiation
time.
VIII Congresso Nazionale di Archeometria
Scienze e Beni Culturali: stato dell’arte e prospettive
Bologna 5 - 7 Febbraio 2014
Fig. 5. Mapping carried out on the area analyzed at different aging times by PLS-DA classification.
Conclusions
Using HSI it has been possible to monitor all degradation phases on the sample. In particular,
in the SWIR range, it has been possible to get information on the non-adherence of the paint
layer to the substrate and the posting occurred during aging in solar box. The use of the HSI
combined to chemometric methods not only allows monitoring the state of alteration over the
time of a painting, but in particular in the SWIR range it is possible to get information before
the deterioration becomes macroscopic.
References
Agresti G., Bonifazi G., Calienno L., Capobianco G., Lo Monaco A., Pelosi C., Picchio R.,
Serranti S.. Surface investigation of photo-degraded wood by colour monitoring, infrared
spectroscopy and hyperspectral imaging, Journal of Spectroscopy. Volume September (2013),
ID 380536
Bevilacqua N., Borgioli L., Adrover Gracia I. 2010. I pigmenti nell’arte dalla preistoria alla
rivoluzione industriale, Casa Editrice Il Prato, 210-213.
Geladi P., Grahn,and J.Burger H.,“Multivariateimages, hyper- spectral imaging: background
and equipment,” in Techniques and Applications of Hyperspectral Image Analysis,
H.Grahnand P. Geladi, Eds., pp. 1–15, John Wiley & Sons, West Sussex, UK, 2007.