Content uploaded by Pierre Drap
Author content
All content in this area was uploaded by Pierre Drap on May 13, 2016
Content may be subject to copyright.
Underwater cartography for archaeology in the VENUS project.
P. Drap, J. Seinturier1, G. Conte, A. Caiti, D. Scaradozzi, S.M. Zanoli 2, P. Gambogi 3
1 LSIS Laboratoire des Sciences de l'Information et des Systèmes
UMR CNRS 6168 – Pierre.Drap@esil.univmed.fr, Julien.Seinturier@esil.univmed.fr
2 ISME - Interuniversity Centre for Integrated Systems for the Marine Environment
c/o DIST, Università di Genova, Italy
gconte@univpm.it, caiti@dsea.unipi.it, d.scaradozzi@univpm.it, szanoli@unian.it
3 MIBAC-SBAT Ministero per i Beni e le Attività Culturali,
Soprintendenza per i Beni Archeologici della Toscana, Firenze, Italy
pamelagambogi@beniculturali.it
Abstract. This article describes a suite of automatic tools to produce underwater georeferenced cartographic data
including archaeologically relevant information. The automatic data processing for marine archaeology goes
from the early data acquisition phase to the building of 3D models of the site and of the objects lying at the site
to the final virtual reality rendering. The paper considers these processes with exclusion of the last phase, and it
describes the applied methodological approach and the obtained results from the Pianosa 2006 mission, which
was performed within the European Project VENUS (Virtual ExploratioN of Underwater Sites). In particular, the
data acquisition system comprises a Remotely Operated Vehicle (ROV) collecting optical data for
photogrammetric processing and georeferenced through an acoustical positioning system. The data are saved on
line in a specific format that makes available the optical image together with the ROV navigation data. The
optical data are processed off-line through standard photogrammetric techniques to obtain a 3D Digital Terrain
Model (DTM). Archaeological artefacts at the site are identified on the DTM, and archaeological-based
information is incorporated in order to produce 3D virtual models of the objects and inserted in the final
representation. Metadata information, including the sequential processing steps to obtain the virtual model of the
artefacts, are stored in a purposely developed data-base system. Evaluation of the results must take into account
two different aspects: the first is the accuracy in the 3D model reconstruction and in the geographical
positioning, measured through standard metric; the second is the evaluation of the archaeologists regarding the
use of the final cartographic instrument.
1 Introduction
The application of most recent ICT (Information & Communication Technology) tools to archaeological field
work has dramatically changed the way in which land archaeologists collect their data, store them and present
their results. Availability of GPS (Global Positioning System) data, progress in optical data acquisition for
photogrammetry and 3D models, integration of geological, geometrical and archaeological information in data
bases with decision support systems, just to name some of the most effective tools, have greatly enhanced the
quality of both the field work and the data from archaeological sites on land. A wide documentation of these
developments, together with references to appropriate projects and results, can be found at the web site of the
European Project EPOCH (Excellence in Processing Open Cultural Heritage) [1]. In deep contrast with the
developments in land archaeology, underwater archaeology field work and data processing is still bound to the
procedures and techniques available at least since the beginning of the ‘90s. The main reason for this lack of
development is that the techniques adopted by land archaeologists are not easily transposed to underwater work:
some of the enabling technologies are simply not directly accessible (as GPS signal), some others are more
problematic to apply (optical data requires specific calibration camera methodologies, as well as way to
compensate random fluctuations in the underwater environment that affect image quality. The gap between ICT
applications in land and underwater archaeology has been recognized in a number of events (as for instance in
[2] and references therein) and several measures have been proposed by national and international Authorities, as
well as by research centres, to bridge the gap. Among these, European Union has selected to fund the project
VENUS: Virtual Exploration of Underwater Sites [3] within the 6th Framework Programme (FP) of Research and
Technological Development (RTD).
The VENUS main goal is to provide underwater archaeologists with a suite of automatic tools to produce
georeferenced cartographic data including archaeologically relevant information and eventually to generate
thorough and exhaustive 3D archives for virtual exploration, to improve the accessibility of underwater sites.
Within the above framework, the project team plans to survey finds and in shipwrecks at various depths and to
explore advanced methods and techniques of data acquisition through Autonomous Underwater or Remotely
Operated Vehicles (AUVs/ROVs) with positioning system and acoustic and photogrammetric equipments.
VENUS research also covers aspects such as data processing and storage, plotting of archaeological artefacts,
Geomatica, 2008, 62(4) pp 419-427
information system management and best practices and procedures for underwater cultural heritage. Further,
VENUS will develop virtual reality and augmented reality tools for the visualization of an immersive interaction
with a digital model of an underwater site. The model will be made accessible online, both as an example of
digital preservation and for demonstrating new facilities of exploration in a safe, cost-effective and pedagogical
environment. The virtual underwater site will provide archaeologists with an improved insight into the data and
the general public with simulated dives to the site.
This paper reports methodologies and results as applied on the first experimental sea trial within the project; the
field test has taken place in Pianosa Island, Tuscan Archipelago, Italy, on October 2006. In this experiment an
ROV has been equipped with video and still camera for photogrammetric data acquisition, and with an acoustic
system able to estimate the ROV absolute georeferenced position. Time stamps are attached to data acquired by
each different system, in order to achieve data synchronization, and, through the definition of an appropriate
format, images are stored together with ROV navigation data. Processing of the data is then performed through
photogrammetry, using also navigation data. Further to the construction of a DTM of the archaeological site,
identified archaeological artefacts on the site (mostly amphorae, in this case) have been modelled geometrically
exploiting specific archaeological knowledge, finally producing an ensemble of cartographic maps summarizing
the DTM at the site, the optical restitution from appropriate mosaic construction of the images, the artefacts both
in terms of geometric models and video images. Additional information, including metadata on the artefacts and
on the site are also made available to the user.
While each piece of equipment used in the system is commercially available, it must be underlined that their
integration and the following exploitation in the processing is not. In fact the key innovative contribution of the
system resides in the strong integration of system and data from the very beginning level of data acquisition;
moreover, feedback from archaeological expertise has always been taken into account at any processing step.
The document is structured as follows. The next section provides a brief description of the archaeological
context; subsequently, survey methods and data processing developed within VENUS are described: calibration,
collection of photographs using ROV and divers, the georeferencing of photographs and amphorae measurement
through photogrammetry using archaeological knowledge; in section 4, data enhanced with 3D information are
reported; finally, current developments in term of a GIS-based operating framework and conclusions are given.
2 The underwater archaeological site of Pianosa Island
Pianosa Island belongs to the Tuscany Archipelago, North Tyrrhenian Sea, Western Mediterranean. The
archaeological site of Pianosa, discovered in 1989 by volunteer divers, is located at a depth of 35 m, close to the
“Scoglio della Scola”, in front of the east coast of the island. The site has been chosen as an operative test-bed
for the VENUS project since its depth allows to survey the area using both robotic equipment and divers. From
previous explorations, the presence of about one hundred amphorae of different origin and epochs has been
documented. The various amphorae range from Dressel 1A (1st century B.C.) to Beltran 2 B and Dressel 20, up
to African models (3rd century A.D.). The Iberian amphorae (Dressel 2-4, Pascual 1, Beltran 2 B) are
predominating and they come either from northern Spain (Tarraconensis) or from southern Spain (Baetica).
The experimental activity has been characterized by a deep interdisciplinary collaboration and consisted in the
collection of georeferenced optical data for photogrammetric reconstruction under the supervision of the
archaeological team of MIBAC-SBAT. The data collection part has been carried out by CNRS with divers and
by ISME with an ROV [4] equipped with a high resolution underwater camera developed by the VENUS partner
COMEX. Prior to the October 2006 experimental activities, MIBAC-SBAT has conducted a cleaning operation
of the site, removing the stratum of dead leaves of posidonia grass from the finds; in addition, MIBAC-SBAT
has conducted a multibeam wide area survey with a RESON 8101 system (240 kHz) operated from a surface
ship with Differential GPS (DGPS). The nominal accuracy of the system at 35m depth is 0.1m.
3 Photogrammetric survey in Pianosa
3.1 Two different ways for data capture
The photogrammetric survey in Pianosa has been planned in order to obtain a sequence of photographs over a
linear strip, with an appropriate forward overlap (60%) between two subsequent photographs. Once a strip has
been completed, an adjacent strip is surveyed, with 20% lateral overlap with respect to the previous strip. The
procedure, as well as the geometry, is very similar to the technique used in aerial photogrammetry; the main
difference is the distance to the seabed and presence of the water. The bathymetric variation could also be in
general an important difference, but not at the Pianosa site, where the seabed in the working area is flat. For the
photogrammetric survey, MIBAC-SBAT operators indicated a 20m x 20m area within the much larger area
previously surveyed by the multibeam system.
Geomatica, 2008, 62(4) pp 419-427
As mentioned, two systems of data acquisition have been tested: the more traditional one, with divers swimming
over the site, and with the robotic equipment. The diver used a Nikon™ D70 digital camera with a 14 mm lens
from Sigma™ and two flashes Subtronic™. The digital camera was embedded in Subal™ housing with a
hemispherical glass. The ROV was equipped with a system provided by the COMEX partner and consisting of a
Nikon DH2 digital camera, a 14 mm lens from Sigma™ and two flashes Nikon™ SB800, with custom-made
housing and connectors (See fig.1). Both divers and ROV kept an altitude of 3m over the seabed during data
acquisition.
Fig. 1. The ROV in the water with digital camera and flash lights in their housing; left: side view of the ROV; right: view from the
seabed upward. The flash lights are on the side of the camera
The working area has been delimited by SBAT divers, that also deployed 4 scale bars (2m each) and a set of 15
markers (concrete blocks 15x15x10cm) in order to define a grid for ROV guidance. The markers delimited zone
was surveyed by the ROV, strip by strip, with the ROV control system keeping automatically fixed the heading
and the altitude over the seabed [5]. With the ROV system, the photogrammetric data were collected in two
different modalities:
- Manually, through a command from the surface ship transmitted via the umbilical cable; a small video
camera installed through the lenses allowed the operator to look at the scene with the same view of the camera
and to issue the acquisition command. In this modality the archaeological expert on board the ship can have full
control of the acquisition operation without actually diving over the site.
- In automatic mode, with a fixed frequency rate, selected taking into account the flash recharge time and
the ROV speed and altitude. An example of two consecutive shots from the ROV automatic modality is reported
in Figure 2.
Fig. 2. Two consecutives photographs from a strip made by the ROV.
3.2 Multimedia photogrammetry calibration
Camera calibration in multimedia photogrammetry is a problem identified since almost 50 years [6], [7]. The
problem has no obvious solution, since the light beam refraction through the different media (water, glass, air)
introduces a refraction error which is impossible to express as a function of the image plane coordinates alone
[8]. Therefore the deviation due to refraction is close to that produced by radial distortion even if radial distortion
and refraction are two physical phenomena of different nature. For this reason, the approach described by Kwon
[9] has been adopted, consisting in the use of standard photogrammetric calibration software to perform the
calibration of the set housing + digital camera. This approach can indeed correct in a large part the refraction
perturbation; however, it is strongly dependent on the optical characteristics of the water/glass interface of the
housing. In order to minimize the refraction error due to this last interface, a housing with a hemispherical glass
(Subal™) has been selected for the divers-operated camera. The same housing, however, could not be
accommodated on the COMEX developed system, employed by the ROV, due to the mechanical constraints
imposed by the additional instrumentation and electronics for the automated mode operation. Hence in this latter
case the housing glass was plane and the refraction action, even after calibration, is much more relevant. A
specific method to compensate separately refraction and distortion has been developed, but its description is
beyond the scope of this paper. The interested reader can find it as a deliverable of the VENUS project
downloadable from the VENUS web site http://www.venus-project.eu.
Comex customized camera
Geomatica, 2008, 62(4) pp 419-427
3.3 The reference system and the ROV navigation data
A fundamental aspect in any survey procedure is the choice of a reference system for the acquired data. The
choice may be driven by the archaeological needs and by the available instrumentation. In general, two choices
can be considered: a relative reference system, and an absolute georeferenced system. The relative reference
system is the option mostly used in underwater photogrammetric work: the reference system is defined from the
data themselves exploiting locally observable geometric features of which the prior orientation and dimension is
known, as buoys to define the vertical axis, scale bars, etc. In most cases this approach requires preparation of
the site with the deployment of appropriate reference objects and tools through divers. This may be a time-
consuming operation, in general not possible beyond diving working depths (about 60m).
Acoustic
links
Umbilical link
Optical Image
(JPEG)
Tele metr y DATA
(EXIF area)
x = East coordinate
y = North coordinate
dist = seabed distance
z = depth
Ω, Φ= pitch and roll angles
K = heading
RPM = thrusters RPM
IMU = Inertial Measurement Unit DATA
USBL = positioning system DATA
Sonar = sonar DATA
Fig. 3. Data flow pipeline for tagging photograph with navigation data.
The use of acoustic localization to track the ROV motion allows to obtain a set of georeferenced positioning data
that can be exploited in the photogrammetric process. In particular, in the Pianosa experiment the ROV was
configured with a USBL acoustic positioning system consisting of a sonar transducer, deployed from the side of
the ship, and a transponder on the ROV frame. The acoustic pings transmitted from the surface ship are reflected
from the transponder; the transducer at the surface can measure the time of flight and the direction of arrival of
the reflected signal, in order to produce an estimate of range and bearing of the ROV from the sonar transducer.
At the surface, this information is merged with DGPS data, taking into account the displacement between the
DGPS receiving antenna and the transducer position. The system used in the experiment (Scout system from
Sonardyne) has a nominal error of 2.7% of the slant range between the surface transducer and the ROV. In
Pianosa, the slant range was always between 50 and 80m, leading to an error between 1.35 and 2.15m. Note that
the DGPS land station to be used for reference has been taken at the same position of the one used for the
multibeam survey. In addition to acoustic positioning, the ROV is also equipped with tiltmeters, compass and
accelerometers: this allows to record the orientation (roll, pitch, heading) of the ROV, hence of the high
resolution camera, taking into account also in this case the geometry of the system, i.e., the displacements
between camera and sensors position. An additional set of sensors is available on the ROV and it is recorded for
post-processing purposes and sanity checks; this set includes depthmeter (pressure gauge), altimeter
(echosounder), encoders on vehicle shafts. Through these combined system, each photograph is associated to the
navigation and system data, in particular with the georeferenced position of the camera and with its orientation
with respect to the seabed, assumed to be flat. Integrated optical/navigation data are directly stored in
JPEG/EXIF format.
One critical aspect in the association of data from different sensors is the synchronization of the system. All the
systems on board the vehicles, including the camera, and the USBL system are synchronized with GPS time at
mission start. GPS clock is always available to all the equipment, through the umbilical cable. However, the
instruments are polled serially before data is written on file, causing a non-constant unknown time jitter. ROV
navigation sensor data are sampled at 30 Hz. Acoustic position is sampled at 1 Hz. Photographs are taken at 0.3
Hz. ROV speed was kept at 0.3 m/s during the experiment, to allow for the necessary overlap between two
consecutive photographs. With the above parameters, and considering the slow dynamic of the ROV vehicle
Geomatica, 2008, 62(4) pp 419-427
with respect to the sampling rate, the error due to the time jitter can be considered negligible with respect to the
other source of errors in the system.
Finally, it has to be remarked that the acoustic positioning system has been also used to track the divers during
their “standard” operation over the field; in this case, however, data association cannot be performed at the stage
of data file writing, but at post-processing stage, relying on data time stamps and on maintaining synchronization
between the diver camera clock and the positioning system throughout the data collection phase. The processing
flow of the referencing system is illustrated in Figure 3.
3.4 Photogrammetric processing
A set of observations of homologous points on photographs are measured manually in order to orient all the
photographs in a local reference system. Then we use the camera position given by ROV navigation data which
are in the absolute reference system consistent with the multibeam data. In Figure 4 an example is given of the
final result: a 3D model of the seabed is obtained, over which oriented photographs are superimposed.
The availability of acoustic bathymetric data from the multibeam survey allows for a comparison with the final
3D model obtained with the photogrammetric approach. It has to be mentioned however that the resolution of the
multibeam survey is of the order of 1 sample every 0.5m, over a large area, while the resolution of the
photogrammetric data is approximately 1 sample every 0.01 m over a much smaller area. The discrepancy
between the DTMs obtained from the multibeam survey and the photogrammetry in the Z direction shows a
mean systematic error of 0.502 m with an RMS of 0.073 m. It is not possible to determine the discrepancy of the
merged data in XY as the seabed is flat in this zone (see Figure 6). This will be estimated from the absolute
accuracy of the measurement techniques.
These discrepancies are indeed compatible with the respective accuracy of each sensor, and in fact may indicate
that the photogrammetric processing does not introduce any additional uncertainty larger than those of the
navigation sensors.
Fig. 4. Oriented photographs visualised in VRML with the non textured mesh of seabed obtained by photogrammetry.
4 Inserting archaeological knowledge: Amphorae plotting
From the oriented photographs, a subsequent 3D geometrical modelling phase of the recorded artifacts
(amphorae in our case) is started. In this phase the modelling must be driven by expert (archaeological)
knowledge; the resulting models, together with the photogrammetric georeferenced data and all the survey data,
is stored in a repository database for further use and interrogation.
The 3D modelling phase procedure consists in exploiting archaeological knowledge to obtain a complete
representation of the measured artefact; it is articulated in two steps:
1) Development of the theoretical model: for each identified object, a geometrical description offers a set
of geometrical primitives, which are the only features to be potentially measured; these are compared
with the theoretical representation of the object as derived from expert knowledge. In our case
archaeologists have identified six amphora typologies, and a theoretical model is produced for each of
them. This theoretical model is formalized in a hybrid way, taxonomy of archaeological artefacts and
XML representation for the amphora typology.
Geomatica, 2008, 62(4) pp 419-427
2) Decision Support System: as photogrammetric measurements are highly incomplete (the object is seen
only partially or may be deteriorated), a rule-based Expert System determines the best strategy to
provide all the geometrical parameters of the studied object, starting from the measurement process and
handling the default data as defined in the archaeological model and the geometrical model. In our case,
we are using the Jess expert system (http://herzberg.ca.sandia.gov/jess/).
The resulting object is thus based on a theoretical model, dimensioned by a photogrammetric measurement. The
modelling procedure is revisable in time, allowing re-processing or complementing the processing as new data
may become available. The whole procedure has been implemented in Java and connected to the Arpenteur
photogrammetric toolbox [10], [11].
Amphorae classification in archaeological work does rely very strictly on dimension information on specific
features of the object, as for instance the neck. In providing a theoretical model for a specific amphora class, it
does make sense to measure these features directly on an available archaeological finds. At the Pianosa site six
amphorae have been resurfaced from the archaeological divers. These amphorae are used as paradigm to define
the theoretical model needed. Since they do not account for all the classes of amphorae observed at the site, the
direct observation of the finds is complemented with drawings and information from archival data; for instance,
type gauloise 3 is modelled accordingly to the typology presented by Archaeological Data Service, University of
York, also partner in the VENUS project [12].
In defining the theoretical model, the diversity of the objects handled by the archaeologists and the geometric
complexity of their surfaces led us to search for stable morphological characteristics of the objects where
diagnostic measurements could be taken. A series of simple geometric primitives are used to approximate these
morphological characteristics and are used as an interface between the photogrammetric measurement and the
underlying model. In the case of amphorae four measurable zones have been defined: rims, handle, belly,
bottom. A set of simple geometrical primitives is fitted by least square method onto the measured points: for
instance a circle on the rim or belly points, a line on bottom point, etc.. This interface allows the user (generally
an archaeologist) to
- Recognize the amphora type on the photographs;
- Choose the amphora type in the interface combo box ;
- Measure a set of points on the zone where measure is allowed;
- Add archaeological comments and observations;
- Insure consistency between observations and theoretical model;
- Store a new instance in the database.
5 Merging results
Bathymetric and photogrammetric data are merged, exploiting the georeferentiation of both acquisition systems
and the orientation adjustments in the photogrammetric processing, and eventually linked to the Amphorae
representation in the database.
Fig. 5. VRML representation of reconstructed amphorae. From left to right: general area bathymetry with the location of the
archaeological site superimposed; a blow up of the experimental site; a blow up on a specific portion of the site, with amphorae
entirely reconstructed using the modelling information. Also visible the measured points on amphorae, and the geometrical
models of a concrete marker (red square) and two scale bars (red and white cylinders).
The textured seabed is obtained by triangulation of the points used to orient the photographs. We have developed
a tool to link each triangle to a set of possible photographs for texturing with the current used photograph
mentioned. The result (3D points, triangle, oriented photographs) are written in XML, X3D and VRML. This
way is very convenient to change the photograph used to texture a triangle or a zone. The database is organized
as a relational database (MySql), and a set of java tools allows to wrap objects from the database and to produce
a VRML representation. The VRML file produced contains a link for every amphora to the database via a PhP
interface. The interface allows the user to see, check and modify the archaeological values regarding the
Geomatica, 2008, 62(4) pp 419-427
amphorae. The user has access to all the data, i.e. measuring points, photos and photo orientation used to
measure the artefact, but these data are read only through the interface (see fig. 6).
6 Current development: GIS framework
In the context of archaeological survey, 2D representations are well known and largely used by archaeologists.
These representations can be handmade drawing or digital maps as used in Computer Aided Design (CAD)
systems or in Geographical Information Systems (GIS).
The complete warehouse managing all the data collected on the site (all 3D information such as seabed DTM,
oriented photographs, 3D artefact reconstruction, ROV navigation, etc.…) allows us to automatically build a GIS
representation of the surveyed site. The 2D GIS representation has two main advantages:
• The 2D representation is convenient for archaeologist needs
• A GIS enable to enhance a simple geometric representation with knowledge
• As we are dealing in the VENUS project with the surface layer only, standard tools for terrestrial 2D
GIS can be relevant.
Such a GIS representation relies on standardized formats: the GeoTIFF [13] and the Shapefile (Shapefile is a
geospatial vector file format from ESRI™ company but it has an open specification and it is used by numerous
GIS software including open source projects). The Shapefile format covers simple 2D geometry representation
and is suitable for a schematic representation of the measured objects. GeoTIFF format enables to store a
georeferenced image.
Fig 6.. 2-D representation with OpenGis platform (geotools/java
development).
Fig. 7. On top representation after photogrammetric
measurement, below automatic 2D representation.
The 2D data exportation is obtained from the 3D representation of the seabed and objects. GeoTIFF and
Shapefile data are independent and automatically generated from the measured objects seabed DTM and oriented
photographs managed by the general warehouse of the project.
The automatic projection constructs an orthophoto from the seabed DTM and oriented photographs following a
given plane. The geographic information is then integrated to the image to generate a GeoTIFF.
7 Conclusions and future work
A solution for underwater survey processing, describing the complete processing flow, from georeferenced data
acquisition in semi-automatic mode to site reconstruction, merging acoustic/optical data with a theoretical model
based on archaeological knowledge, has been presented. In the framework of the VENUS project a work is in
progress to define an ontology for underwater archaeology and more precisely for amphorae present on any
given site [14].
In addition of the site survey presented here, VENUS consortium plans to immerse archaeologists inside a virtual
universe depicting a reconstructed archaeological site, for example a shipwreck, and allow them to work on this
site as naturally as possible. The digital model generated by the survey will then be used, with the help of virtual
reality and mixed reality, for constructing immersive, virtual environments that enable archaeologists and
general public to experience an accurate and fully immersive visualization of the site.
Geomatica, 2008, 62(4) pp 419-427
8 Acknowledgements
The divers Giuseppe Adriani and Paolo Vaccari discovered the Pianosa underwater archaeological site in 1989.
Paolo Bonaiuti and Emiliano Africano, assisted in the cleaning site operations and in several stages of the
experimental test. Venus partners COMEX S.A., Marseille, France, and ADS, Archaeological Data Service,
University of York, provided equipment and documentation for part of the work described here
The Venus sea trial operations in Pianosa, October 2006, have been made possible thanks to the voluntary
support of a number of different Institutions: Corpo Nazionale dei Vigili del Fuoco - Direzione Regionale Vigili
del Fuoco della Toscana - Nuclei Sommozzatori (Diving team of the Tuscany Fire Brigade); Italian Ministries of
Justice, of Transportation and Navigation, of the Environment. Cooperativa Ormeggiatori Piombino, Studio
Archeologico Thetys, Geosystem Parma.
VENUS is partially supported by the European Community under project VENUS (Contract IST-034924) of the
"Information Society Technologies (IST) programme of the 6th FP for RTD". The authors are solely responsible
for the content of this paper. It does not represent the opinion of the European Community, and the European
Community is not responsible for any use that might be made of data appearing therein.
9 References
1. EPOCH: European Network of Excellence in Open Cultural Heritage, (last visited July 2008),
www.epoch-net.org
2. G.Vettori, A.Rizzerio, P.Gambogi, G.Casalino, A.Caiti, Recent experiences in underwater archaeology
surveys with remotely operated vehicles in the North Tyrrhenian Sea, in T.Akal, R.Ballard, G.F.Bass (Eds.),
Proc. Int. Conf. on The application of recent advances in underwater detection and survey techniques to
underwater archaeology, Bodrum, Turkey, May 2004.
3. P. Chapman ,G. Conte ,P. Drap ,P. Gambogi, F. Gauch ,K. Hanke ,L. Long ,V. Loureiro ,O. Papini ,A.
Pascoal ,J. Richards ,D. Roussel. VENUS, Virtual ExploratioN of Underwater Sites. in VAST/CIPA. 2006.
Nicosia, Cyprus: Archaeolingua, Budapest, Hungary.
4. G. Conte, S. Zanoli, D. Scaradozzi, L. Gambella. Robotics tools for underwater archaeology. in
ECSAC07 VII International Conference on Science, Arts and Culture. Science for Cultural Heritage. 2006. Veli
Lošinj, Croatia.
5. G. Conte, S. Zanoli, D. Scaradozzi, L. Gambella, A.Caiti. Data gathering in underwater archaeology by
means of a remotely operated vehicle. in XXI CIPA international Symposium. 2007. Athens, Greece: The CIPA
/ VAST, Nicosia, Cyprus.
6. Georges F. Bass and Donald Rosencrantz. L'utilisation des submersibles pour les recherches et la
cartographie photogrammétrique sous-marine. L'archéologie subaquatique, une discipline naissante. 1973. Paris.
7. ASP, Manual of photogrammetry, Fourth Edition. 1980: American Society of Photogrammetry (ASPRS
Pub.) ISBN 0-937294-01-2.
8. Maas, H.-G. New developments in Multimedia Photogrammetry. in Optical 3-D Measurement
Techniques III. 1995: Wichmann Verlag, Karlsruhe.
9. Kwon, Y.-H. Refraction at the Water-Air Interface. 1998 [cited; Available from:
http://kwon3d.com/theory/dlt/refr.html.
10. Pierre Drap, Julien Seinturier, and Luc Long. A photogrammetric process driven by an Expert System:
A new approach for underwater archaeological surveying applied to the 'Grand Ribaud F' Etruscan wreck. in
Applications of Computer Vision in Archaeology ACVA'03 2003. Monona Terrace Convention Center,
Madison, Wisconsin, USA.
11. Pierre Drap and Luc Long, Photogrammétrie et archéologie sous-marine profonde. Le cas de l'épave
étrusque Grand Ribaud F XYZ, 2005. N°103, partie 1 et N° 104, partie 2.
12. ADS, Archaeology Data Service, University of York, UK, Website.
//ads.ahds.ac.uk/catalogue/archive/amphora_ahrb_2005/details.cfm?id=135.
13. Niles Ritter and Mike Ruth. GeoTIFF 1.0 specification by the GeoTIFF working group at
http://www.remotesensing.org/geotiff/spec/geotiffhome.html
14. Robert Jeansoulin and Odile Papini. Underwater archaeological knowledge analysis and representation
in the VENUS project: a preliminary draft. in XXI CIPA international Symposium. 2007. Athens, Greece.
Geomatica, 2008, 62(4) pp 419-427
