MENTS
the monument
r this reason,
s currently an
lish a suitable
detail without
upporting any
ys have to be
on Walls. The
points without
input, but the
close range
letails that are
inuities in the
neters for the
ests are based
] reliability of
:edure: firstly,
ing have been
ue to the non
1ophotos as à
ical buildings,
> in such areas
( any work is
"uction of the
lost effective
chniques; the
g information
Baratin et al.,
re, colours or
or differential
0 a horizontal
Bitelli, Gabrielle
plane which represent an element of or the entire digital surface model of the object; image resampling is carried out
using different kinds of transformations (Kraus, 1997).
The orthophoto support is given by the object DSM, which can be created with various manual or automatic methods.
Close range photogrammetry has been proven to provide significant cost savings over conventional measurement
techniques. However, the photogrammetric processing stage has typically been slowed due to reliance on manual
procedures. Introducing automation to photogrammetric processing is an obviously desirable advantage. However,
varying scale, large depth of field, obstructions, kind of object surface and lighting differences can pose difficulties in
automating photogrammetric processing in close range environments. The procedure followed for generation of the
DSM directly influences the quality, in terms of precision, of the orthophoto produced. For better results, a preset
procedure must be followed, considering that even if automatic DSM generation is currently a most interesting practice
for replacing a repetitive and expensive stage of conventional photogrammetric production, its level of accuracy is not
always clear (Achilli et al., 1998; Bitelli et al., 1999-a; Zanutta, 2000).
In the research described, which takes as its example an 18th-century gate in the fortification walls on the island of
Malta (Zabbar Gate, at Cospicua; figure 1), a low cost digital photogrammetric station (StereoView Menci Software)
was used to carry out a series of tests designed to identify the most suitable matching parameters for automatic
generation of the DSM of the monument. Following generation of the DSM, several tests were applied to define the
optimum procedure for construction of digital orthophotos. This optimised procedure would be used not only to identify
the correct parameters which guarantee metric precision in the product obtained, but also a level of quality which makes
it a valid support for the various analysis operations involved in the restoration of a monument: quality of the materials,
analysis of their deterioration, etc.
The research is part of a project for the recovery of the system of fortifications on Malta, which arose as a result of co-
operation between the University of Bologna and the Maltese government (Baratin et al., 1998).
2 SURVEYING THE FACADE OF THE ZABBAR GATE
The Zabbar Gate (figure 1) was built after 1706 by the Roman architect Fortunato Carapecchia. The gate’s design is one
of a vast series of “ideas for great churches” and “military defence works” for Malta which Carapecchia perfected
during his stay on the island. For this gate, a survey of the facade was carried out: firstly, artificial control points (retro-
reflecting signals) juxtaposed with the walls were topographically measured, then the photographs were taken and,
finally, 1:100 vector plotting of the facade was carried out.
-b-
Figure 1 — a) Map of the Malta Island; b) Zabbar Gate façade along the Cottonera Lines.
The topographical survey was realised by a Leica TC2000 total station. The space resection method was used starting
from two stations located at the same distance from the façade, with measurement of the azimuthal and zenithal angles
and, where possible, inclined distances. The least squares adjustment of the network produced the object coordinates of
the signalised control points with error ellipses (95% confidence level) of just a few millimetres.
The nadiral photogrammetric images, acquired using a Wild P31 metric camera with lifting truck at various heights,
were taken in such a way as to cover the entire object with sufficient stereoscopic overlap. The shooting distance of
approximately 20 m gave the photograms a scale of 1:200.
The plotting, carried out using the Galileo Siscam Digicart 40 analytical plotter, produced conventional graphical and
numeric results such as prospects, sections and profiles in a scale of 1:50, suitable for restoration of the structure (Bitelli
et al, 1999-b).
The structure examined is entirely represented by four stereoscopic models obtained from orientation of the nadiral or
psuedonadiral photograms. In order to generate an image of the façade with metric characteristics easily manageable in
a CAD environment, the errors linked to the use of simple perspective rectification of a photogram were evaluated. The
object surveyed has a maximum overhang of around 2.5 metres in the balcony area. In the models there are various
shadow zones in which the stereoscopy is null or of poor quality. Generation of a pure perspective rectification would
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000. 63