|
Chapter
3.2 3D Scanning
of Artworks Lica Pezzati, Raffaella Fontana CNR – INOA, Istituto Nazionale di Ottica Applicata, Firenze,
Italia
Contents 3.2.2
Single-Line Laser Scanning 3.2.2.1
Case Study 1: The Minerva of Arezzo
3.2.3.2
Case
Study 4: Perugino’s Paintings
3.2.3.3
Case
Study 5: Greek and Roman Coins Following the lines discussed in the previous Chapter 3.1, in this section we
extend the review of 3D scanning techniques currently adopted in the
diagnostics of artworks to 3D scanning of objects: statues, paintings,
archaeological findings, and other small- and medium-scale pieces of
historic and cultural interest. Optical techniques play an important role in
the field of Cultural Heritage diagnostics, because of their safety and
effectiveness. Among them, 3D scanning techniques have recently found
the way to a variety of relevant applications. They are presently used
to measure buildings, statues, inscriptions, coins, and even paintings,
whose apparent “flatness” attracted little
attention from the pioneers of 3D measuring techniques. The growing
technological progress of laser-based devices opened a new era in 3D
survey, enabling the design of highly accurate instruments for quota
measurements and allowing for extremely dense data sampling at high
acquisition rate. The relative ease of acquisition of 3D arrays of
several million points fostered the study of problems related to the
representation and use of digital models. The uses for these models
range now from historical and artistic studies to the assembling of
easy-access and long-lasting digital archives, as well as from the
analysis of the conservation condition to the monitoring of the
restoration treatments. Furthermore, the digital model constructed with
3D data can be used for virtual reality applications. The work of art
can be enjoyed outside its habitual surroundings and the image can be
placed in different contexts. Any change resulting from a proposed
restoration treatment can be visualised (so-called computer aided
restoration) as well as the proposed reconstruction of losses can be
visually evaluated. Moreover, by measuring the shape over time it is
possible to obtain data on the object alterations. This allows
mechanical stresses to be located, the effects of microclimatic
variations (temperature and/or humidity) to be quantified, surface
degradation or the monitoring of shape variations introduced by the
restoration process to be measured. Different 3D scanning techniques are generally
based on different principles and capture different and often
complementary information. As described in Chapter 3.1, the most common
techniques used for 3D measuring are the optical triangulation,
embedded in most of 3D measuring system, and the time-of-flight, but
other means, such as interferometric
methods (and among these, holography), were and are used. As a
reference, see in Fig. 3.2.1 a classification of the more common 3D
measuring technique, adopting the schematic division between active and
passive methods, where active means that the measured object must be
actively illuminated, while passive methods work with ambient light.
Another useful classification is the one based
on the object scale: not all these techniques can indeed be used for
all possible objects, if optimal results are to be attained. For
instance, where time-of-flight (TOF) methods are well-suited for large
and distant objects, like architectures and terrains, fringe projection
is very good for medium-scale semi-flat objects with smooth continuous
surfaces, such as paintings and inscriptions, single-line laser
scanning is ideal for medium-scale objects like statues and vases, and
finally micro-profilometry
is well suited for highly-detailed nearly-flat surfaces, either of
small objects or of big object details. In this Chapter we present two prototype
instruments designed and assembled at INOA for medium-scale and
small-scale 3D acquisition, and a variety of their applications to the
diagnostics of Cultural Heritage objects. For a general introduction on
the use of 3D measuring methods and 3D data processing, see at the
previous Chapter 3.1.
3.2.2 Single-Line Laser Scanning A laser-line scanner is a device for 3D
measurements based on optical triangulation through single-line laser
projection. It combines resolution with versatility, making it
particularly suitable for in situ applications both
under difficult lighting conditions and in all environments. A custom
laser-line scanner has been designed at INOA for scanning medium-scale
objects. The prototype (Fig. 3.2.2) is composed of a
diode-laser line-projector emitting red light (λ = 670 nm) shaped to a light sheet by an
optical element (prism) and a HRES CCD camera (1300 x 1030 pixel) placed at a certain, known, angle
with respect to the laser source. The camera is equipped with a 16 mm
lens and a band-pass filter, chosen to match the wavelength of the
laser in order to eliminate problems related to ambient light. A
motorized stage allows for scanning the objects and closely-spaced
parallel profiles are then acquired.
The system was calibrated and corrections were
applied to compensate for lens distortions. However, the accuracy of
the image spot location, and hence the instrument depth accuracy, is
degraded by laser speckle. With a stand-off distance of 50 cm and a
scanned area of about 30 x 30 cm2, the quota
resolution resulted to be 50 μm, with an absolute error smaller than 0.3 mm.
3.2.2.1 Case
Study 1: The Minerva of The Minerva of For the realization of the digital model of the
statue, 119 separate measurements, or range maps,
were acquired, each with a spatial density of about 16 points (vertex)
per mm2, and each framing a different part of
the statue surface. The
range maps were then joined (registered)
and merged
in a single 3D model (mesh),
that is shown Fig.
3.2.3b.
The 3D model of the Minerva stands as the only
documentation of the statue’s form before the restoration
process irreversibly changed it. For instance, during the
statue’s repair the wooden inferior part
as well as the right arm were removed, and the corroded
surface has been smoothed by cleaning. The 3D model will be used either
to reproduce and place exactly the missing parts, or to keep trace of
possible further alterations of the shape.¸
The Minerva restoration of 1785 appears to be
incorrect. The recovery circumstances are not well-known, as well as
the original preservation conditions of the statue that most probably
had suffered a previous restoration in the XV century. The pose of the
right arm, that puts the Minerva on an oratorical move, probably
imitates the one of the Arringatore,
another major bronze statue of the Tuscan museum, together with the
Chimera and the Idolino.
A previous plaster repair, whose evidence is given by a printing dated
the beginning of XVIII century, describes the right arm of the Minerva,
goddess of war and wisdom, laid out along the trunk and the forearm up
to hold a lance: this is a recurrent attribute for the warrior goddess,
armed with the Corinthian helmet. The digital model allowed the
realization of a software application capable to visualize the two
hypothetical positions for the statue’s right arm (Fig.
3.2.5).
3.2.2.2
Case Study 2: Poesia
Cave Inscriptions The Grotta
della Poesia (literally: The Cultural Heritage Department of the Near the cave, in the extended archaeological
site of
A very accurate device, capable of measuring
distances in the micrometric range, is necessary to reveal the finest
details of an object surface. The prototype instrument realized at INOA
for micro-profilometry
makes use of a distance-meter device based on an interferometric technique (conoscopic holography). This
method combines the capabilities of working on different materials, of
measuring reflective and diffusive surfaces, and of being insensitive
to colour changes. The probe, shown in Fig. 3.2.8 is a video camera
coupled with a conoscopic
module, consisting in a special optical crystal (uniaxial
birefringent)
sandwiched between two circular polarizers.
A laser diode (λ = 680 nm) is focussed to a point on the object
surface; the conoscopic
module splits the light beam reflected from the investigated point, in
two beams (the ordinary and the extraordinary beams), which due to an
interference phenomenon produce a patterns of fringes, recorded by the
video camera. The distance of the investigated point from the probe is
then obtained by measuring the fringe spacing. The conoscopic micro-profilometer (Conoprobe, Optimet) used in the prototype
is equipped with a 50 mm lens. This set up results in a dynamic range
of 8 mm at a working distance of 40 mm. The out-of-plane resolution is
about 1 mm, the accuracy is better than 6 m,
and the in-plane lateral resolution is about 20 μm. The probe is mounted on two motorized linear
stages, allowing a maximum scanning area of 280 x 280 mm2. The
acquisition speed ranges from 100 to 400 points/sec, depending on the
spatial sampling step and the acquisition area chosen. This scanning micro-profilometer
was applied on a variety of objects, ranging from panel paintings to
details of statues, with different instrument configurations purposely
designed. The methods can reveal details which are not visible at sight
and often not detectable with other techniques. Due to its very high resolution, the instrument
stillness during measurements is crucial. Vibrations in the measurement
environment can cause spurious details to appear in the acquired
surface model.
High-density surface sampling with micrometric
resolution, besides providing the users with an accurate reproduction
of the surface micro-features, provides a data set that can be used for
statistical calculations such as roughness computation. Roughness
measurement of an artwork is important to document the surface
condition, to assess either changes due to restoration or surface decay
due to wearing agents and to monitor the time evolution of the shape.
Michelangelo’s David (Fig. 3.2.9a),
exposed at the Museo dell’Accademia in Roughness measurements of statues is quite an unexplored field, due to the lack of
measurement protocols or reference case studies. Roughness is defined
in terms of deviations from a mean surface level and it is usually
described by the RMS (Root Mean Square) value around it (Rq)
and the wavelength (lq). Therefore Rq
is a measure of the roughness amplitude and lq is a measure of its characteristic length. For
roughness computation, we selected 1 x 1 cm2
locally flat areas. To correct for any minor shape variations
(waviness) within the sample areas, for each acquired line a quadratic
fit was performed and the result subtracted from the data. The RMS
roughness value was computed for each acquired line. Its mean value and
standard deviation were then finally calculated for each sampling area.
The results are plotted in Fig. 3.2.12, together with the error bars
representing the measurement variability range.
3.2.3.2 Case Study 4: Perugino’s Paintings The ‘Canaa
weddings’ panel painting (Fig. 3.2.13a) is part of a
‘predella’
by Cosimo Vannucci, called il Perugino,
from the Galleria Nazionale
dell’Umbria
in Perugia (Italy). We performed a survey of the panel to analyse
some peculiar carvings, apparently originated by toothed tools used in
laying the preparation. Measurement results are shown in Figs. 3.2.13b
and c. The sizes of two investigated areas are respectively 26
x 15 cm2 and 15 x 10 cm2 and the sampling
density is 64 points/mm2. The three-dimensional
model is displayed as a raking-light image. Parallel marks are visible
as traces of toothed tools, having different depth and spacing. Fig.
3.2.14 shows the acquisition software interface. Numerical data can be
obtained by the surface profiles: the maximum depth of the traces is 50
μm and the spacing ranges from 0.5 to 1.5 mm.
The integration of the 3D relief with a colour
image is shown in Figs. 3.2.13d and e for the two investigated areas.
The integration of the colour information with the 3D model increases
the readability of the data, thus making the analysis easier for the
art historian or the restorer that is studying the object.
3.2.3.3
Case Study 5: Greek and Roman Coins Application of 3D techniques to archaeology
faces two main problems, often shared by other fields but particularly
critical in this case: the first is the huge variety (dimensional,
chromatic, of materials and shapes) of the objects, ranging from
buildings to skeletons, from wreckages to statues, etc. The second is a
complete lack of generally accepted measurement standards for
innovative methods such as 3D surveys. Archaeological findings are
usually documented by means of photographs and drawings, thus relying
on the photographer and on the illustrator skill. These two methods are
scarcely descriptive of the real shapes and sizes of objects, because
they are based on plane representations. 3D modelling being an
objective means for measuring and recording an object shape represents
a big step up. The information contained in a 3D digital model largely
surpass that stored in a photograph, so 3D techniques are very
promising for archaeological documentation, and new standards for this
must be studied. We present an application of the optical micro-profilometry to two bronze
coins, kept at the Cultural Heritage Department of the The first coin (Figs. 3.2.15a and b), from
In Fig. 3.2.16 a few examples of image
processing are shown, where an embossing filter was used for obtaining
raking light simulations. On the obverse side of the Greek coin (Fig.
3.2.16a) a human head is clearly visible: Apollo, according to the
archaeologist’s opinion: On its reverse side a lyre and the
inscription kapu are
marked (Fig. 3.2.16b). Due to the highly worn out surface of the Roman
coin, interpretation of the results was difficult. On the obverse side,
the two-faced Janus
head can be hardly seen (Fig. 3.2.16d), whereas on the other side a
prow of a ship is probably present (Fig. 3.2.16e). The two images of
the drawing of the coin type are also shown (Figs. 3.2.16c and f).
For the numismatist the primary source of
knowledge of a coin history is contained in the inscriptions. The
quantitative morphological characterization of coin surface and the
subsequent possibility of image processing is, thus, a new approach
that could completely change the way to document these findings.
Non-invasive optical techniques are
particularly suitable for Cultural Heritage diagnostic applications. In
particular, 3D scanning of objects and of surfaces has recently found
the way to a variety of relevant applications, because it allows fast
measurements of the shape of artworks with both high accuracy and high
resolution. 3D scanning techniques, together with new modelling
software tools, allow a high fidelity reproduction of artworks that can
be applied either to support and
document its repair, or for the realization of 3D archives and virtual
museums. Moreover, starting from a high-resolution digital model of an
object, a further step could be its reproduction by means of
fast-prototyping techniques, like stereo-lithography or
electro-erosion. In this Chapter we presented two laser-scanner
prototypes designed and assembled at INOA for medium-scale and
small-scale 3D acquisition: a single-line laser scanner and a conoscopic micro-profilometer. The former has a
scanning surface 30 x 30 cm2 wide with a stand-off
distance of about 50 cm and a resulting depth resolution of 50 mm. Its hardware simplicity and daylight working
capabilities, and, above all, the capability to acquire range maps of
very complex 3D objects, make it useful for a variety of Cultural
Heritage applications. The latter allows measurements on a maximun area of about 28
x 28 cm2 with a stand-off
distance of about 4 cm, a maximum transversal resolution of 20 μm and a 1 μm quota resolution. The high-density surface
sampling with micrometric quota resolution, besides a high fidelity
reproduction of the surface characteristics, provides a data set that
can be used for statistical calculations. A few applications of both devices to the
diagnostics of Cultural Heritage objects were shown. 3D data sets were
used either for creating digital models of the artworks, or for
studying the finest surface characteristics by means of both image
analysis and roughness computation.
J.F. Asmus:
More light for art conservation. IEEE Circuit
and Devices Magazine, March 1986 p.6. A. Cabrucci:
La scansione laser nel rilevamento architettonico, Quaderni di restauro
– Tecniche di rilevamento nuove frontiere della tecnologia, ed. E.S.I,
Napoli, 157/2001. C. Rocchini, P. Cignoni, C. Montani, P. Pingi, R. Scopigno: A Suite of Tools for
the Management of 3D scanned Data. Proceedings of the 2001 Workshop of
Italy-Canada on 3D Digital Imaging and Modelling Applications of
Heritage, Industry, Medicine & Land, Padua, Italy, 3-4 April
2001. S.F. El-Hakim: A practical approach to creating
precise and detailed 3D models from single and multiple views.
Proceedings of the XIX ISPRS Congress, P. Cignoni,
C. Montani, C. Rocchini, R. Scopigno: External Memory
Management and Simplification of Huge Meshes, IEI-CNR Tech. Rep., Jan.
2001, pp.10. B. Curless,
M. Levoy: A volumetric
method for building complex models from images. SIGGRAPH ’96
Proceedings, pp. 303-312, ACM
SIGGRAPH, 1996. M. Garland, P.S. Heckbert:
Surface simplification using quadric error metrics. In: SIGGRAPH 97
Conference Proceedings, ACM SIGGRAPH, 1997, pages 209 –216. M. Levoy,
K. Pulli, B. Curless, S. Rusinkiewicz: The Digital
Michelangelo Project: 3D scanning of large statues. Comp. Graph. Proc.,
Annual Conf. Series (Siggraph’
00), pp. 131-144, Addison Wesley, H. Rushmeier,
F. Bernardini, J. Mittleman, G. Taubin: Acquiring input for
rendering at appropriate levels of detail: digitizing a Pieta. Eurographics Rendering Workshop
1998, G. Drettakis and
N. Max (ed.), Springer Wien,
1998. M. Gaiani,
M. Balzani, F. Uccelli:
Reshaping the Coliseum in Rome. Computer Graphics Forum, Vol. 19(3), pp
369-378, 2000. R. Baribeau,
M. Rioux: Influence of
speckle on laser range finders. Appl.
Opt. Vol. 30 No. 20, pp 2873-2878, July 1991. R. Baribeau,
M. Rioux, G. Godin: Recent Advances in the
Use of a Laser Scanner in the Examination of Paintings. Restoration
'92, V. Tood et al.
(Editors), United Kingdom Institute for Conservation, United Kingdom,
pp. 69-73, NRC 35040 (1992). K. Pulli:
Multiview
registration for large datasets. Proc. Second Int. Conf. on 3D Digital
Imaging and Modeling,
Ottawa, Canada, October 1999, pp.160-168. C. Rocchini,
P. Cignoni, C. Montani, R. Scopigno: The Marching
Intersections Algorithm for merging range images. IEI-CNR Tech. Rep.,
submitted paper, July 2000. M. Garland, P. S. Heckbert:
Surface simplification using quadric error metrics. SIGGRAPH 97
Conference Proceedings, ACM SIGGRAPH, 1997, pp. 209-216. R. Fontana, M.C. Gambino,
M. Greco, E. Pampaloni,
L. Pezzati, R. Scopigno: High-resolution 3D digital models
of artworks. Proc.
SPIE. 2003. V. 5146. p.
34-43. R. Fontana, M.C. Gambino,
M. Greco, L. Marras, M. Materazzi, E. Pampaloni, L. Pezzati: The diagnostics of statues: a high-precision surface analysis
of roughness of Michelangelo’s David. Proc.
SPIE. 2003. V. 5146. p.
236-243. R. Fontana, M.C. Gambino,
M. Greco, L. Marras, M. Materazzi, E. Pampaloni, L. Pezzati, P. Poggi: Integrating 2D and 3D data for diagnostics
of panel paintings.
Proc. SPIE V. 5146, 88 (2003). D.J. Whitehouse: Handbook of Surface Metrology.
UNI-ISO 468, Rugosità delle superfici
– Parametri, loro valori e regole per la descrizione delle
caratteristiche (1991). G.Y. Sirat,
D. Psaltis: Conoscopic holograms. Opt. Comm.
9(65), 243-245, 1988. G.Y. Sirat,
D. Psaltis: Conoscopic holography. Optics
Letters 10(1), 4-6, 1985. J. F. Asmus:
Holograph interferometry
of painted surfaces. Proc. Int. Cong. on ‘Conservation of
Wood, in Painting and the Decorative Arts’, Oxford, United
Kingdom, 1978, Vol. IIC, 141-144. D. Charlot et al.:
Holographie conoscopique
– reconstructions numeriques.
Annales des Telecommunications
43 (1988) 23-26. D. Bertani,
M. Cetica, M. Melozzi, L. Pezzati: High-resolution
optical topography applied to ancient paintings
diagnostics. Optical Engineering 34/4 (1995) 1219-1225. Y.Y. Hung: A new optical method for strain
measurement and nondestructive
testing. Opt. C. Lahanier,
G. Aitken, J. Taylor,
F. Blais, L. Borgeat, J.-A. Beraldin, M. Rioux, B. Breuckmann, P. Colantoni, L. Cournoyer, C. De Deyne, M. Picard, R. Pillay, G. Godin: The 2D multi-spectral
digitisation and the 3D modelling of easel paintings. to be presented, ICOM-CC, 14th Triennal Meeting ICOM-CC, 12-16
Sept. 2005, J. Taylor, J.-A. Beraldin,
G. Godin, L. Cournoyer, R. Baribeau, F. Blais, M. Rioux, J. Domey: NRC 3D Technology for
Museum and Heritage Applications. The Journal of Visualization and
Computer Animation 14/3 (2003).
CNR – INOA, Gruppo Beni Culturali: http://arte.ino.it For more links see Chapter
3.1.
Luca Pezzati CNR – INOA Istituto Nazionale di
Ottica Applicata Largo E. Fermi 6 I - 50125 Firenze Italia E: luca@ino.it Raffaella Fontana CNR – INOA Istituto Nazionale di
Ottica Applicata Largo E. Fermi 6 I - 50125 Firenze Italia E: lella@ino.it |
