24-06-2014, 02:34 PM
Quantitative Hyperspectral Reflectance Imaging
Abstract
Hyperspectral imaging is a non-destructive optical analysis technique that can
for instance be used to obtain information from cultural heritage objects unavailable with
conventional colour or multi-spectral photography. This technique can be used to
distinguish and recognize materials, to enhance the visibility of faint or obscured features,
to detect signs of degradation and study the effect of environmental conditions on the
object. We describe the basic concept, working principles, construction and performance of
a laboratory instrument specifically developed for the analysis of historical documents. The
instrument measures calibrated spectral reflectance images at 70 wavelengths ranging from
365 to 1100 nm (near-ultraviolet, visible and near-infrared). By using a wavelength tunable
narrow-bandwidth light-source, the light energy used to illuminate the measured object is
minimal, so that any light-induced degradation can be excluded. Basic analysis of the
hyperspectral data includes a qualitative comparison of the spectral images and the
extraction of quantitative data such as mean spectral reflectance curves and statistical
information from user-defined regions-of-interest. More sophisticated mathematical feature
extraction and classification techniques can be used to map areas on the document, where
different types of ink had been applied or where one ink shows various degrees of
degradation. The developed quantitative hyperspectral imager is currently in use by the
Nationaal Archief (National Archives of The Netherlands) to study degradation effects of
artificial samples and original documents, exposed in their permanent exhibition area or
stored in their deposit rooms.
Introduction
Spectral imaging refers to the acquisition of a series of digital images at a number of different, welldefined
optical wavelengths. Traditionally, if the number of wavelength bands is smaller than ten, the
term Multispectral Imaging (MSI) is used. MSI is used extensively in many application areas,
including the field of restoration and conservation of artworks, where MSI can be regarded as common
technology (see e.g. [1, 2]).
If the number of wavelength bands is much larger than ten, usually the term Hyperspectral Imaging
(HSI) is used. HSI has already proven its worth in various fields such as agricultural research,
environmental studies and defence (e.g. [3, 4]). HSI instruments used for these applications typically
are mounted on an aircraft or satellite to record the surface of the Earth. A recent example is the
Hyperion HSI instrument, which is part of the EO-1 satellite that monitors the Earth's surface [5, 6].
On a microscopic scale, HSI technology is steadily becoming a valued research tool, especially in
biomedical research (see e.g. [7, 8]). In these cases, the HSI instrument is often combined with an
optical microscope [9].
However, it is also possible to use HSI for investigation of objects of cultural heritage. In the last
few years this technique has been introduced for the imaging of artworks [10-15]. Recently, Kubik
described how a portable HSI instrument based on a digital camera system in combination with
narrow-band optical filters can be constructed and used for distinguishing pigments on paintings
Measurement principles of a HSI instrument
Implementation concepts of hyperspectral data acquisition
Before developing a HSI instrument for a particular range of applications, and making a detailed
system design and component selection, it is very worthwhile to compare the various implementation
concepts at a rather abstract level. Each concept has a number of principle strengths and weaknesses,
and their impact on the particular application at hand should be taken into account.
2.1.1. Assumptions about acquired hyperspectral data
In the following, we assume that a hyperspectral data set of an object contains three dimensions of
information, namely two spatial dimensions and one spectral dimension. This means that we exclude
the relatively few cases, where either only one spatial dimension is required, or where spectral data is
acquired from all three spatial dimensions of an object. By considering only a single spectral
dimension, we exclude ultraviolet fluorescence and infrared luminescence spectral imaging, whichQHSI
QHSI instrument design
The QHSI instrument is based on two identical wavelength tunable light sources (nick-named
TULIPS A and TULIPS B) and a monochromatic CCD camera, as shown schematically in Figure 5.
The object (e.g. a historic document, a piece of textile, etc.) under investigation is illuminated by these
two TULIPS under an angle of 45º with respect to the object plane and imaged by the camera under an
angle of 0º from above. This illumination geometry was chosen, because it is known to result in a good
approximation of an ideal diffuse illumination where each location on the object is evenly illuminated
from all directions
Analysis and visualization methods
The analysis of a hyperspectral measurement depends of course very much on the application at
hand. Nevertheless, browsing of the calibrated spectral reflectance images of the object and their
qualitative comparison is often a very effective means of obtaining information about otherwise
invisible features. This is the same approach as used with conventional multi-spectral imaging devices.
However, due to the much increased number of spectral bands and the very homogeneous response
over the entire FOV, the hyperspectral images may often provide somewhat more information.
Figure 13 shows as an example a real-colour and some hyperspectral images of the Map of Syracuse
(drawn circa 1680), which is part of the Admiral M. de Ruyter Fond of the Nationaal Archief in the
Hague [30]. Due to corrosion of the drawing ink, neither the real-colour image (covering the visible
range from 380 to 780 nm) nor the individual spectral images (spectral bandwidth 10 nm) taken at
400 nm and 600 nm, respectively, show all the fine details of the drawing. This means that the
reflectance in this spectral range is very similar for the areas of the original ink lines and for the
corroded areas. However, the optical properties for these two regions differ substantially in the nearinfrared
wavelength range. The corroded areas become transparent in this wavelength range, so that the
higher reflectivity of the paper substrate itself provides a high contrast with the ink lines. For example,
in the 800 nm infrared image it is possible to discriminate clearly between the original ink lines and the
corroded areas. In this way, the skill and effort of the artist applied to work out the fine details of the
Summary and conclusions
In summary, we report on the development of a prototype instrument for hyperspectral reflectance
imaging of historic documents. The instrument records spectral reflectance images at 70 different
wavelength bands (bandwidth is typically 10 nm) ranging from 365 to 1,100 nm. As opposed to
conventional multi-spectral imaging devices, the recorded data are calibrated by using reference
measurements of a Spectralon® standard reflectance target. The resulting hyperspectral datacube
therefore provides a set of 70 spectral reflectance values for each of the 4 million locations (pixels)
measured on the object.
The measurement concept was chosen to be the most suitable for measuring delicate stationary
cultural heritage objects that can be exposed to only a very limited amount of light without risking
degradation. The specific design of the realized instrument prototype concerning the spatial resolution
and the measured area are optimized for the analysis of historical documents, which requires for
example retrieving valid reflectance data also from within the thin ink lines of handwriting. We discuss