CIPA 2003 XIX"' International Symposium, 30 September-04 October, 2003, Antalya, Turkey 
vision algorithms, which process images taken of a calibration 
object. The adopted Tsai camera model (Tsai, 1987) uses 5 
parameters (focal length, centre of project, pixel aspect ratio and 
1st degree radial distortion). Range information was used to 
estimate the external parameters (the translation and rotation of 
the camera relative to the global reference frame). 
Finally per scan triangulation, based on the 2D grid provided by 
the scanner (Turk, 92), was created. The main problem with per 
scan triangulation is that the meshes from different scans need 
to be merged after triangulation, which can be a difficult task. 
nThe final result of laser data processing was a textured 3D 
model which can be measured and compare with model. 
7.2 Photogrammetric data 
The photogrammetric process, described in section 4, provides a 
set of geometrical data on the lacked geometry coming from 
laser scanner measurement. The lacked geometry has two 
different origins: the object is broken (a part of the lips is 
destroyed) and a artificial hole was simulates in the belly part to 
represent a physical obstacle during the laser scanning phase. 
The photogrammetry gave us the photograph orientation, in the 
same reference system, a textured mesh obtain with an 
automatic measurement process and data to generate theoretical 
model. 
8. MEASURING OBJECT WITH A THEOTERICAL 
MODEL 
8.1 Elaboration of a Specific Body 
The measurement process proposed in this work is relying upon 
the hypothesis of the existence a theoretical model of the 
architectural objects studied. We can suggest a theoretical 
model for the amphora founded in the Etruscan wreck. 
All of these amphorae are of Etruscan origin and have the same 
shape as amphorae individualized by A. Py in 1974 (type Py 4) 
in a study of the imports to Vaunage and Villevielle (Gard). 
Eleven years later, Gras and Slaska completed this initial 
classification by proposing a typology of amphora from 
Southern Etruria. The type Py 4 and its variants have been 
included in the EMD group (for detail see [Py M. et F., 1974] 
and the work of Sourisseau, 1997). This regularity in the 
production of the amphora allows us to use a modelling 
approach and to formalize this knowledge into a hierarchy of 
objects sharing the same properties and structured according to 
the Object paradigm. The amphorae from the wreck, for the 
time being, have been grouped into four sub-classes of the Py 4 
type amphorae according to morphological considerations. 
8.2 The use of an expert system 
The objects heterogeneity handled by the archaeologist and the 
geometric complexity of their surfaces led us to search for 
stable morphological characteristics of the objects where 
diagnostic measurements could be taken. These diagnostic 
characteristics are also described in the model. 
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. The measurement can have two purposes: object 
orientation and position and determine their intrinsic 
characteristics. 
Fig. 4. The Expert System general schema. Fig.5. Ideal amphora model: from archaeologist design to digital model 
The photogrammetric measurement is supported by some 
strategic points on the amphora (Fig. 4, on the right side). Five 
areas are used to redundantly define the amphora coordinate 
system definition parameters. If measurements to certain parts 
of the amphora are impossible, the coordinate system 
determination mechanism uses relationships between amphorae 
(if existent) or default values. The inference problem of values 
relying upon incomplete data or data that needs to be re 
evaluated is frequent in archaeology. 
Obtaining an amphora 3D representation requires a merge of 
photogrammetric measurements (S2) and theoretical model 
(SI). The technique that we use consists in supplementing 
photogrammetric measurements by the theoretical model. These 
two data sources give information on the remarkable zones of 
the amphorae, defined by an archaeologist (lip, belly, handles, 
back, see Fig. 4). It is obvious that the difficulty of the source 
merge consists in finding at which time to use a source or the 
other. The solution used by our system is the recourse to 
propositional logic to formalize the knowledge contained in the 
two sources and how to use it. The merging method is provided 
by an inference on the logic formalization of the sources which 
generates actions to be achieved. Finally the merge is carried 
out according to the actions. [Drap, Seinturier, Long, 2003].