-14- 
directly on the square of the distance between the camera and 
the object. A high precision is therefore achieved by suitably 
limiting this distance, and thus the illuminated area. The 
acquisition of relatively large objects, as a statue of human size, 
requires a large number (typically hundreds) of partial views (or 
range maps), taken all around the object, that have to be 
integrated through a complex alignment procedure in order to 
represent the whole surface. Using portable equipment, 
employed to scan Heritage artworks kept in museums, such 
range images are taken from unknown locations, and the 
coordinate system of each point cloud is referred to the range 
camera. In order to refer all the acquisitions to the same 
coordinate system, adjacent range maps are acquired with 
partial superimposition (at least 30%), and rototranslated by 
means of special software algorithms that minimize the mean 
square deviation or find a best match between overlapping 
points of adjacent maps (Besl, McKay, 1992; Chen, Medioni, 
1992). 
2.2 TOF based 3D scanning 
Another class of 3D acquisition systems - often indicated in the 
photogrammetry world with the term “laser scanner” - is based 
on the simultaneous measurement of the time of flight of light 
pulses sent towards the object, giving the distance between the 
scanner and the measured surface, and the vertical and 
horizontal angles. The kind of output is similar to that 
attainable with a triangulation scanner. 
The main advantage is that long-range imaging is possible, thus 
allowing large structures to be imaged. However, the achievable 
accuracy is relatively low, never better than a few millimeters, 
but commonly resulting in the centimeter range. Also in this 
case a procedure of partial views alignment is required. 
2.3 Digital Photogrammetry 
Photogrammetry is a well consolidated technique, that has a 
long history of developments for over a century. Digital 
photogrammetry is the product of an intensive research 
conducted for the last 20 years for the automation of 
information extraction from digital images, based on image 
analysis methods. 
Photogrammetry is based on the acquisition of multiple images 
of the object to be measured, taken from point of views 
sufficiently apart by a photo camera. The method requires the 
selection of recognizable elements on the acquired images, 
including reference targets of a known size. Once the selection 
of homologous points has been performed, the 2D coordinates 
on the photogram plane are processed using an algorithm of 
bundle adjustment that allows the corresponding 3D coordinates 
to be estimated. Photogrammetry results generally suitable to 
large scale measurements. Close-range photogrammetry cannot 
be used for camera-object distance lower than 1-2 meters. The 
process of the identification of the points to be measured, being 
manual or semi-automatic, requires a long a tedious work if the 
number of points is elevated. 
The available technology is largely used in the metric survey of 
architectures and archaeological sites (El-Hakim et al., 1996). A 
unique feature of photogrammetric techniques is its extremely 
high accuracy, that ranges from 100 ppm to 10 ppm of the 
overall dimension of the imaged area. 3 
3. CASE STUDY: BAPTISTERY OF FLORENCE 
The object of our study was a Romanic mosaic and part of the 
surrounding area, located in the underground of the Baptistery 
of Florence, at the north-west comer. 
The measurement conditions in such environment are 
particularly difficult: the bad conservation state after the early 
dislocation to a museum of the most valuable pieces (1912-15, 
1925) and the difficult access don’t make easy the monument 
understanding. 
Surveys based on overall measurements and inaccurate 
photography’s compositions have been made up to 1990, but 
due to the complexity of the archaeological site such kind of 
measurements and are prone to artificial regularizations, giving 
possible fake results. The first systematic studies over these 
structures were started in 1990 and published later (Marino et 
al., 1996). 
The mixing of mosaic portions with wall structures of different 
ages makes difficult the historical interpretation of the artifact. 
In this case the differences in the pavement construction 
techniques, such as stones shape and cutting methods or mortar 
characteristics and thickness, are the only elements for an 
appropriate reading of the monument. 
In the examined room a portion of a wall adjacent to the mosaic 
is semicircular and is sheltered by the foundation walls of the 
apse. All the room walls are made with three different layers: 
the lower one has a regular structure, the intermediate - less 
homogeneous - shows slid connections, while the upper one is 
mainly made by conglomerate. 
On the roof lie a mosaic fragment representing geometrical 
drawings and made by small black and white tesserae, of the 
roman period. 
4. METHODS AND INSTRUMENTATIONS 
A well established praxis in studying an artifact like the one 
previously described is to make topographic surveys integrated 
with photographs and 2-D drawings (generally 1:10 scale) 
showing elevations. For the smaller details a contact relief on 
polyvinyl is generally used, where each mosaic tesserae or the 
overall mosaic drawing are taken. In this way important mosaic 
details are acquired. From these raw data some papers are 
worked out, such as thematic maps of the materials employed, 
of the gaps, of restorations, of the conservation state, etc. 
A tridimensional relief is often acquired through molds made in 
silicon or glass-reinforced plastic in order to accurately evaluate 
the presence of roof depressions. 
The introduction of advanced 3D measurement technologies 
seemed therefore very advantageous for eliminating all the 
manual steps typically employed for this kind of survey. 
4.1 Topographic survey 
11 points have been identified on the mosaic surface, and 
marked with reflecting targets. Their topographic coordinates 
have been measured with a steel band for planimetry, and with 
geometric leveling performed with a digital level meter Leica 
NA3003 and invar stadia bar for the points height. 
4.2 Digital photogrammetry 
Photogrammetry in this application act as the “global reference” 
supplying a single coordinate system for repositioning piece of 
data coming from other sensors. Such global reference is then 
reoriented to the absolute reference defined by older surveys. 
In order to have a good level of redundancy for measurements 
control, it was decided to make photogrammetry with two 
different kind of equipment. 
4.2.2 Semimetric camera: A semimetric camera Rollei 6006 
(6x6) was employed with a calibrated lens with 40 mm focal 
lenght. Each photogram, characterized by an average scale of 
1:40, was then acquired with a calibrated Mirage scanner 
capable of 1400 DPI.