|
Chapter 3.1 3D Laser Scanning of Architectural Sites Margarida Pires1, Claude
Borg2 1DOP, INETI, Lisboa,
Portugal
Contents
3.1.2.1 Principles
of Laser Triangulation: Information Acquisition
3.1.2.2 Principles
of Laser Time of Flight: Information Acquisition
3.1.2.4 Information
Processing 3.1.3 Methodology
and Sase Studies
3.1.3.1 Case
study 1: Rupestrian Art
in Coa Valley (UNESCO no.866)
3.1.3.2 Case
study 2: Roman Ruins in Lisbon
3.1.3.3 Case
study 3: Manoel Island
Chapel,
3.1.3.4 Case
study 4: Kordin Temples
III,
3.1.3.5 Case
study 5: Tas-Salvatur
Chapel, Historic sites and cultural heritage are
subjected not only to natural ageing but also to natural accidents,
inappropriate interventions and modifications and vandalism. That is
the reason why artworks documentation is so important, mainly in what
concerns fragile artefacts, monuments to be restored, or archaeological
sites under excavation works. Moreover the documentation of artworks,
especially high accuracy documents, allow for the construction of real
and virtual models, providing basis for restoration studies,
dissemination, and animation. Artworks documentation using optical techniques
is based mostly on the registration of light reflected from the
objects, no matter these objects are illuminated by the sun light or a
camera flash light. Depending on the sensor used in the recording
device, the stored information might be only the intensity of light
received from each point of the object e. g. black and white
photographic film or monochromatic CCD cameras, or both intensity and
colour, such as colour photos or infra-red thermal images. Due to the
incoherent nature of usual light sources, these techniques are used to
obtain 2D (two-dimensional) images of the objects, this is, records
that contain information and are reproducible in a plane or flat
surface. Photogrametry
could be considered an exception, because this technique allows the
construction of non flat records of objects; but photogrametry is based on the
acquisition with different viewing angles, of plane images of the
object and it is the post-processing and combination of these 2D
records that allow obtaining volumetric information of the object. However, in our world the material objects or
artworks have volume, no matter we are considering the architecture of
a monument façade, the shape of a ceramic pot or the brush
stokes in a painting or the inscriptions on a paper document. So, the
objects must be recorded and documented in 3D (three-dimensional or
volumetric) images in order to contain more complete geometrical
information and to make possible volumetric reproductions, like
material replicas or virtual images. 3D Laser Scanning is a technique that takes
profit of the coherence properties of laser light, consisting in a very
pure colour and highly directional light beam, to acquire, store and
process 3D computer images and information of objects, using a low
power laser beam as the light source and detecting the light reflected
from the object surface on very sensitive sensors. Scanning the laser
light across or around the object surface allows for a complete
three-dimensional record of the surface. Laser scanners offer a sophisticated
alternative to the conventional data collection methods, with very good
resolution and accuracy. Laser scanning equipment is lightweight and
easy to operate in open field. Because the scanned laser light
doesn’t cause any change on the object as a contacless technique, it can be
used in precious or fragile objects with different shapes and sizes,
and the 3D record can be used later and remotely, for study,
documentation, demonstration and reproduction, with no more handling of
the real object.
3D laser scanning techniques might use three
different basic operation principles: — The first one uses the interference
phenomena of laser light and the analysis of the optical fringes
produced, to detect and image an object surface texture and profile.
It’s usually applied to small areas (smaller than 1 cm2)
and can achieve high resolution (less than 1 micrometer). This
technique, also called micro-profilometry,
is mainly used in micro technologies but not usually found in artworks
documentation, so it won’t be detailed further. — The second operation principle is
called laser triangulation, in which mathematical relations between the
direction of the emitted laser beam and the direction of the detected
reflection allow for information on the position of the object surface
points. This technique is applied to objects with typical dimensions
from 1 cm to 1 m and can achieve a few micrometers
resolution. It is usually used for museological
artworks as well as for details on selected parts of larger monuments. — The third basis for operation of
this technique is called laser time of flight, and using the known
value of the speed of light in air, detects the distance to a point on
the object surface by measuring the time interval between the emission
of a laser pulse and the detection of its reflection. The scanning of the laser beam over the
complete surface of the object is obtained in different manners,
associated to the dimensions of the imaged areas and the required
resolution. In micro-profilometry
is usually the small object that is moved on a high precision two axes
translation stage. Laser triangulation uses often not a laser beam but
a laser line on the object that is swept along the object surface and
usually several scans in different positions are needed in order to
register all the object surface (top, bottom, sides), although if
appropriate the object can also be moved on a rotating plate. In laser
time of flight the emitted laser pulse is angular scanned along two
perpendicular directions, illuminating a matrix of points on the object
surface.
3.1.2.1 Principles of Laser Triangulation: Information Acquisition Let’s consider a linear laser beam
incident on an irregular surface (Fig. 3.1.1). In the point (P,
P’), where the laser beam strikes the surface, a luminous dot
can be observed, no mater the angular position of the observer. This is
due to the diffuse reflection, meaning in many directions, of the laser
light from the object surface. However, the optical system (O), usually
a lens and some filters used to collect the reflected light and
transmit to the detector only “views” a very narrow
reflected beam, due to its reduced aperture (Fig. 3.1.1a). The angle (α) between the directions of the
incident and reflected beams decreases when increasing the distance (D)
of the illuminated point on the surface to a reference point on the
equipment. Yet, the optical system (O) collecting the reflected light
from the surface point produces, like a photographic camera, a point
image (I) on the detector surface (D), having a position that depends
on the angle α and so on the distance of the illuminated point (P)
(Fig. 3.1.1b).The painting was examined by the double exposure
technique in stable laboratory environment with controlled thermal
excitation and surface monitoring. The random high reflectivity of the
varnish layer caused localized contrast loss obstructing the capturing
of a normalized intensity photographic image. The interest operator
notes that the overall uniform bright and dark fringe formation -or
fringe pattern- due to its overall uniform response to the temperature
alteration is locally interrupted by smaller fringe patterns generated
by various subsurface discontinuities. The next step is to locate and
isolate by zooming in to those discontinuities in order to extract
those patterns so that at the final step only defects are seen, some
characteristic ones are shown in Fig. 3.4.3.
The detectors used for laser triangulation
consist usually of CCDs
(charge coupled devices) made of a mosaic of individual photo sensitive
cells- the pixels. Each pixel on the detector surface receiving light
due to the image formation of the point of an object,
produces a small current, used to identify the position of the image
point on the detector. Using mathematical relationships between the
triangle LPO (Fig. 3.1.1a) formed by incident beam ↔ reflected beam ↔
optics to laser distance, and the triangle CC’I (Fig. 3.1.1b)
formed by optical aperture centre (c) ↔ sensor centre (c’) ↔
image point (I, I’), the equipment software computes the
distance of the illuminated point to the reference point based on the
position of the detector cell receiving the light of the image point.
Each point of the object surface is defined by a set of three values or
coordinates, relatives to a reference point pre-defined by the system. The two other coordinates of the illuminated
point are dependent on the direction of the emitted laser beam, which
is controlled and known by a central controller. PSD (position sensing
devices) sensors might also be used to give information about the image
point position, but they are not so commonly found. In order to have complete information on the
object surface it is necessary to acquire information of a multitude of
points in the object surface, this is the position of these points
defined by the three coordinates of each one. To achieve this, a
mechanical or opto-mechanical
system, providing translation, rotation or deflection is used to scan
the laser beam that illuminates the selected points on the object
surface, and get the position information about each one and so about
the surface geometry. To achieve a faster acquisition and storage of
the information related to the huge number of required points, most
laser triangulation 3D scanners emit not a laser beam illuminating a
single point, but a laser sheet or curtain, illuminating a contour line
on the object surface (Fig. 3.1.2).
Due to the fact that laser sources are
monochromatic, the large majority of 3D laser scanners don’t
register information about surface colour, but only the shape and
luminance of the object surface. However some systems exist emitting
polychromatic light or in conjunction with extra colour sensors,
allowing for the simultaneous acquisition of colour and shape of the
objects. The number and spacing of points or contour
lines needed to reproduce the object surface or in other words the
resolution of the acquisition process is mostly defined by the
operator, taking in mind the detail required, the surface geometry and
the total area to be covered. It is also limited by the specifications
of the equipment used. Obviously the higher the resolution and number
of points imaged, more time consuming is the acquisition process and
more heavy the files storing the surface information.
3.1.2.2
Principles of Laser Time of Flight: Information Acquisition For longer ranges of artwork dimensions, such
as monumental façades, historic buildings or archaeological
sites, time of flight scanners or also called ranging scanners are
used. A short laser pulse is sent to the object surface, reflected back
and detected on a photosensor.
Both the pulse emission time and the returning energy detection time
are registered by a central controller; the time between the emission
of the pulse and the detection of the reflection can be measured. Using the known speed of light in the air and
the laser pulse travel time since emitted by the laser source till the
detection on the sensor, the distance to the illuminated surface point
may be easily computed.
The scanning system sweeps the object surface
to be imaged in several parallel lines (Fig. 3.1.3), while the system
laser head emits laser pulses at high repetition rates, and detects the
back-reflected energy and hence computing the distance of the
illuminated points. The remaining process is (in principle) quite
similar to the previous description. The direction of the emitted
laser, known by the central controller, defines the other two
coordinates of each point, in respect to the system reference point,
and information about the surface geometry results from the position of
a multitude of points in the object surface, defined by the position or
coordinates of each one. A complete 3D model usually comprises of a
number of scans from different viewpoints. Special identifiable targets
are required to consolidate (join) all the scans from the different
viewpoints. These targets can be easily detected by the scanning
software. The most common targets include spheres and plane targets
with high reflectivity usually supplied by the manufacturer.
The equipment used for 3D laser scanning has a
central controller unit, usually installed in a transportable computer,
besides the laser head (laser source and detector), and some kind of
stable support (e. g. a tripod or a solid scaffolding) with a centrally
controlled movement system, in order to perform the beam scanning. The multitude of object surface points detected
by the sensor are digitized in the sensor controller and stored in the
computer. The central controller software processes the data sent by
the sensor and converts the data into the coordinates of each point and
their correct place in respect to each other, forming a
“point cloud” that is then processed by the
software to edit and display the 3D shape obtained by each scan on the
object surface (Fig. 3.1.4). The central controller unit also controls and
receives information on the position of the laser head, having in this
way information on the relative position of the successively
illuminated points or lines of the object surface. Moreover the
software of the 3D laser scanning equipment allows for input
information filtering, image parallax correction, elementary images
consolidation and a few basic operations on the 3D image, such as
magnification, rotation, translation, or others, depending on the
equipment being used. These basic operations should be used during
field works for information acquisition and storage, in order to check
that a complete set of scans has been performed over the surface to be
documented.
3.1.2.4 Information Processing Most usually the information or image
processing capabilities of the software provided with 3D laser scanning
equipment are quite limited, and a more appropriated commercial image
processing software is used in order to have further manipulation of
the stored data, to join all the set of scans of the object surface and
get more information about the artwork object geometry. The image processing software opens the files
produced by the equipment software and has to be worked out then with
the adequate format. The first operation done (automatically by the
software) to the imported files is to consolidate (join) and align the
several records obtained in the same area of the object surface,
obtaining the 3D image of that area. To aid the manipulation of the
data during this phase, automatic systems (a mesh of small triangles),
or geometrically formed primitives (cube, sphere, cone etc) are
projected to the respective points. The software modules will have a
considerable influence on the quality and on the time needed to achieve
a good result. A user friendly efficient software product is a basic
requirement. The several partial images, also called mosaic
or tiles are merged, taking into account the overlapping of the scans
and the viewing angle in which each scanned image was obtained. After
that, the editing of the complete 3D images or models is possible,
usually with the insertion of cross lines that make easier the 3D
visualisation, and also other surface effects such as surface texturing
or colour.
3.1.3 Methodology and Case Studies Usually the basic methodology applied is quite
standard, as listed below. However the detailed procedure required may
differ from case to case, and depends obviously on the Artwork to be
documented, namely size and detailed features. The 3D laser scanning
equipment and image processing software to be used, and the final
objective of the task will identify a particular methodology. The
equipment will also determine the maximum applied resolution and the
number of necessary viewpoints. Methodology: — Acquisition planning –
after an art object detailed analysis, its necessary to define the
total area to be imaged, the required resolution, the distance between
laser head and object surface, the area imaged in each scan for the
defined resolution and the stand off distance, the total number of
scans needed to cover all the area and, in different angles in order to
eliminate shadow areas and surround completely the object surface (if
required), the scans overlap for multiple scans and the position of the
equipment in each scan, according with the previous definitions. — Equipment positioning and fixing in
a rotation or translation very stable device, facing the artwork
surface at the defined stand-off distance. — Equipment calibration, as defined
by the equipment manufacturer, after the equipment is settled. — Start laser scanning and
simultaneous data storage, area by area, with relative displacement of
the equipment (when needed). — Merging the partial areas when
multi scans are used in one total image — Image analysis and detection of
defects, such as shadow areas, occlusions, too much bright areas, or
accidental lack of overlap. Possible additional scan data if necessary.
— Storage of partial images and file
backup — 3D image(s) post processing, such
as aligning, merging, meshing, texture mapping and editing.
3.1.3.1 Case Study 1: Rupestrian Art in Coa Balley (UNESCO No. 866) Collection of rupestrian
rock engravings from Palaeolithic Superior (22 000 to 10 000 BC), one
of the first manifestations of Human Art (http://whc.unesco.org/en/list/866).
The small width and depth of the lines engraved
in the rocks required a high resolution data acquisition (sub
millimetre); and to improve the resolution, the stand off distance
between the laser head and the rock surface needed to be decreased
(less than 0.5 m). This results in a decreased field of view. However
the area of each engraving was too large to be covered by only one
scan. So, each one of the several engravings has been divided in
several partial areas, called tiles (Fig. 3.1.5). The work plan
included the identification of the area, number and overlap of tiles
needed to obtain a good resolution file of each rock engraving. Using the post image processing software at the
main office, the tiles were joined, aligned and merged, and the 3D
Image, containing all dimensional information was edited (Fig. 3.1.6)
and used for reproductions available in the local museum.
3.1.3.2
Case Study
2: Roman Ruins in Lisbon Water drainage conducts, made of bricks and
mortar, part of sub-soil Roman constructions of 1st century AC, under
Emperor Augustus.
In this case the object to be documented was
formed by several surfaces forming sharp angles between each others, so
several scans were needed from distinct points of view, in order to
cover the entire 3D surface without shadow areas (Fig. 3.1.7). After
merging, the editing of the 3D image is possible, usually with the
insertion of lines that make easier the 3D visualisation (Fig. 3.1.8).
In order to improve the 3D images edited, the
several commercial software have usually functions allowing noise
filtration, artificial colour modelling of the 3D images obtain
sections and profiles of the surface and take also measurements (Fig.
3.1.9).
3.1.3.3
Case Study
3: Manoel Island St. Antonio Chapel, in Fort Manoel Island was built by the
Knights of Malta between 1723 and 1740, under the patronage of
Portuguese Grand Master Manoel
de Vilhena. During the
Second World War the chapel was severely damaged by aerial bombing and
it's currently under restoration. The aim of this task was to produce a 3D model
of the remains of the chapel (Fig. 3.1.10) to assist the conservators
and conservation architects to be able to understand each plane and
study an ideal material for the 'reconstruction' of the chapel. The
facade was scanned from three different viewpoints. The sides and
interior were scanned from further six viewpoints. Although the church is partly covered with
scaffolding, the laser software allows the possibility of 'removing'
the scaffolding (Figs. 3.1.11 and 3.1.12). An older photograph without
scaffolding was used to texture map the facade.
3.1.3.4
Case Study
4: Kordin Temples III, On the Kordin
heights above The instruments used included a total station
which in this case produced the first cloud of points, a laser scanner
using time of flight AND another laser head based on laser
triangulation. An external camera was used for texture mapping.
Fig. 3.1.14a shows the cloud of points of one
'apse', in which each colour identifies a different viewpoint. Fig.
3.1.14b shows the data by the total station, the time of flight scanner
the laser triangulation data merged together for Kordin Temples III.
3.1.3.5
Case Study
5: Tas-Salvatur Tas-Salvatur Chapel (Fig. 3.1.15) is one of Kalkara's earliest buildings of
note. In 1670, the Bailiff, Fra
Giovanni Bichi, nephew
of Pope Alexander VII, decided to build a country villa on the
peninsula in front of the church.
On his death during the epidemic plague of 1676, Fra Giovanni Bichi, was buried in the church
of Tas-Salvatur. The
church is presently undergoing intensive conservation and restoration
treatment. It was necessary to produce a 3D model of the
church from the inside and outside, to assist the conservators and
conservation architects to be able to understand the state of
deterioration and deterioration mechanisms in relation to each other.
The 3D model was to serve as a fundamental document too. Two laser scanners were used, a time of flight
AND a laser triangulation system. The scans of both scanners were
integrated to form one whole. The facade was scanned from seven
different viewpoints with the time of flight scanner. The sides and
interior were scanned from a further six viewpoints with the laser
triangulation scanning system.
The various scanning viewpoints of the time of
flight scanner and laser triangulation scanner produce different images
represented in different colours (Fig. 3.1.16, left image) that are
consolidated together, producing the cloud of points, visualised in
this way (Fig. 3.1.16, right image) when the luminance map of the
object is superposed to the cloud of points. With image post-processing, various sections/
plans may be achieved from the 3D model. This example (Fig. 3.1.17)
shows the thickness of the walls, in the apse and cupola.
The altar and effigy were scanned in a very
tight grid using a laser head that works by triangulation. Selective
use of the objects was made and the data of the two scanners combined
to form a whole 3D image of the altar (Fig. 3.1.18).
3D Laser scanning is safe because it employs
eye-safe laser systems, because the principle of operation relies on
the object selective illumination and laser reflection detection with
no further interaction with the artwork and yet because, unlike
photographic methods it’s completely environmentally safe. 3D Laser scanning is versatile, because can be
used with many different materials, except for absolute transparent or
reflective objects, in different environments, underwater inclusive,
and for a large range of objects size, from archaeological sites, to
very shallow and almost invisible inscriptions on stone. 3D Laser scanning can achieve accuracies of few
micrometers, in objects of several meters size, allowing for the
detection of very small features in the artwork surface and a very
detailed documentation of the object. However the high accuracy of the 3D Laser
scanning technology has also associated inconvenient, consisting in the
high data content of the image records, being time consuming and
requiring high capacity and high speed computers for the image post
processing operations.
Remondino, F., Guarnieri,
A., Vettore, A.: 3D
modelling of close-range objects: Photogrametry
or laser scanning? Proc. of SPIE – 5665 (2005), 216-225. Picard, A., El-Hakim, S., Godin,
G., Valzano, V., Bandiera, A.: ''Combining 3D
technologies for cultural heritage interpretation and entertainment'',
Proc. of SPIE – 5665 (2005), 108-118. Giunta, G., Di
Paola, E., Castiglione,
B. ''Innovative 3D information system for the restoration and the
preventive maintenance plan of the Fontana, R., Gambino,
M., Mazzotta, C.,
Greco, M., Pampaloni,
E., Pezzati, L.
''High-resolution 3D survey of artworks'', Proc. of SPIE –
5457 (2004), 719-726. Fontana, R., Gambino,
M., Gianfrate, G.,
Greco, M., Pampaloni,
E., Pezzati, L. ''A 3D
scanning device for architectural survey based on time-of-flight
technology'', Proc. of SPIE – 5457 (2004), 393-400. Scopigno, R., Cignoni,
P., Callieri, M., Ganovelli, F., Impoco, G., Pingi, P., Ponchio, F. ''Using optically
scanned 3D data in the restoration of Michelangelo's David'', Proc. of
SPIE – 5146 (2003), 44-53. Ricci, R., Fantoni,
R., Ferri De Collibus, M., Fornetti, G., Guarneri, M., Poggi, C. ''High resolution
laser radar for 3D imaging in artwork cataloguing, reproduction and
restoration'', Proc. of SPIE – 5146 (2003), 62-73. Fontana, R., Gambino,
M., Gianfrate, G.,
Greco, M., Marras, L., Materazzi, M., Pampaloni, E., Pezzati, L. ''Time of flight
laser scanner for architectural and archaeological applications'',
Proc. of SPIE – 5146 (2003), 185-193. Fontana, R., Gambino,
M., Greco, M., Pampaloni,
E., Pezzati, L., Scopigno, R. ''High-resolution
3D digital models of artworks'', Proc. of SPIE – 5146 (2003),
34-43. Toylor, J., Beraldin,
J.-A., Godin, G., Cournoyer, L., Baribeau, R., Blais, F., Rioux, M., Domey, J. ''NRC 3D imaging
technology for museum and heritage applications'', J. Visualization and
Computer Animation, 14 (2003), 121-138. Bordone, A., Ferri
De Collibus, M., Fantoni, R., Fornetti, G., Guarneri, M., Poggi, C., Ricci, R.
''Development of a high resolution laser radar for 3D imaging in
artwork cataloguing'', Proc. of SPIE – 5131 (2002), 244-248. Fontana, R., Greco, M., Materazzi,
M., Pampaloni, E.,
Pezzati, L., Rocchini,
C., Scopigno, R.
‘’Three-dimensional
modelling of statues: The Minerva of Arezzo’’, J.
Cultural Heritage, 3
(2002), 325-331. Forest, J., Salvi,
J. ''A review of laser scanning three-dimensional digitisers'', IEEE
Int. Conf. Intelligent Robots and Systems, 1 (2002), 73-78. Afshar, M., Boroumand,
M., Studnicka, N.
''Archaeological scanning of Levoy, M., Rusinkiewicz,
S., Ginzton, M.,
Ginsberg, J., Pulli,
K., Koller, D.,
Anderson, S., (...), Fulk,
D. ''The digital Michelangelo project: 3D scanning of large statues'',
Proc. ACM SIGGRAPH Conf. Computer Graphics (2000), 131-144. Axelsson, P. ''Processing of laser scanner data -
Algorithms and applications'', ISPRS J. Photogrammetry
and Remote Sensing, 54 (1999), 138-147. Beraldin, J.-A., Blais,
F., Cournoyer, L., Rioux, M., Bernier, F., Costa, M. "Surface Inspection by Optical
Triangulation Method", Optical
3D Laser Mapping – Services: http://www.3dlasermapping.com Archaeoptics – Services: http://www.archaeoptics.co.uk Ciber F/X – Services: http://cyberfx.com/art_sculpt I Site: http://www.isite3d.com Inus Technology Inc – Software RapidForm 2001: http://www.rapidform.com InnovMetric – Software Polyworks:
http://www.innovmetric.com/ Laser Design – Equipment and Services
& Engineering: http://www.laserdesign.com Leica Geosystems
– Equipment and Services: http://hds.leica-geosystems.com Mensi – Equipment and Servicest: http://www.mensi.com MetricVision – Equipment: http://www.metricvision.com Metrologic Group – Software Metrolog: http://www.metrologic.fr/uk/ Nutfield Technologies – Equipment: http://www.nutfieldtech.com Optech – Equipment: http://www.optech.ca/prodilris) Polhemus – Equipment: http://www.polhemus.com/fastscan Quanta Point – Services: http://www.quantapoint.com Raindrop Geomagic
– Software Geomagic
Studio: http://www.geomagic.com RSI – Equipment and Services: http://www.rsi.gmbh.de/hls_e.htm ScanSite – Services: http://www.scansite.com/art Shape Grabber– Equipment: http://www.shapegrabber.com Trimble – Equipment: http://www.trimble.com/3dlaserscanners Various: http://www.commerce-database.com/laser-scanners Z+F UK Ltd – Software Light Form
Modeller: http://www.zf-uk.com
3D Scanning Software, an introduction: http://www.i3mainz.fh-mainz.de/publicat/korfu/p11_Boehler.pdf Heritage 3D.org: http://www.ceg.ncl.ac.uk/heritage3d/downloads/TLS%20formats%20V1.pdf i 3 Mainz:
http://www.scanning.fh-mainz.de Laser Scanning and Modelling, Industrial and
Architectural Applications: http://www.commission5.isprs.org/wg4/workshop_ancona/proceedings/56.pdf Leica Articles: http://hds.leica-geosystems.com/articles/articles.html Terrestrial Laser Scanning- Applications in
Cultural Heritage and Civil http://www.commission5.isprs.org/3darch05/pdf/19.pdf The Effects of Reflecting surface Materials
Properties on Time of Flight Laser Scanner measurements: http://www.isprs.org/commission4/proceedings/pdfpapers/180.pdf Rock Art: Past and Present: http://www.dur.ac.uk/prehistoric.art Stonehenge Laser Scan:
http://www.stonehengelaserscan.org The Digital Michelangelo Project: http://graphics.stanford.edu/projects/mich
Margarida Pires DOP INETI E. Paço do Lumiar 22 PT - 1649-038 Lisboa Portugal Claude
Borg Heritage Malta Royal Naval H. Bighi Kalkara CSP 12 Malta |
