CI P A 2003 XIX th International Symposium, 30 September - 04 October, 2003, Antalya, Turkey
The segmentation work was carried out on a part of the
building which contains the Department of Georesources and
Land of the Politecnico di Torino. The survey was performed
using a RIEGL LNS-Z420 laser scanner. Apart from the
façade, the laser scanner also surveyed some internal parts of
the building and part of the external pavimentation.
It was hoped to obtain the main plains of the object using the
developed software: the façade, the external pavimentation
and then the remaining parts.
The result was positive. Only a few minutes were necessary
to carry out the whole operation.
From the figure it is possible to see how the objects
belonging to different plains were in fact grouped together in
different layers. These are visualised in the image by
assigning different colours to them. The model can be
exported into AutoCAD environment, assigning a different
layer to each entity.
3.2 Region growing and Analysis of the main components
A scansion of the “loggetta” of Sansovina was used for the
evaluation of the segmentation capacity of the region
growing model and for the analysis of the main components;
this — can be found on the lower part of the bell tower in
Piazza San Marco in Venice.
He survey was carried out using a RIEGL LMS-Z420 laser
scanner.
Figure 5 - the “loggetta” of Sansovina - Venice - Italy
This is an interesting archaeological example as the object
has a series of complex structures, including columns, some
internal niches inside which there are statues and bas-reliefs.
Two different types of elaborations were carried out.
The software that was used for the elaboration combines
region growing techniques with an analysis of the main
components (Roggero, 2002). Some describers were defined
for the segmentation (inertia moments and tensors, curvature
and static moments).
Limits were imposed on the curvature in order to be able to
correctly carry out the aggregation.
A first elaboration of the data was carried out using not very
rigid constraints on the aggregation. This means that during
the segmentation procedure the most evident discontinuities
were evaluated while the small discontinuities were not
considered.
The results of this first elaboration are shown in figure 6
The segmentation algorithm identified 14 main portions of
point clouds (the columns, the statue, portions of walls etc.).
The software that was used for the elaboration organised the
results in an AutoCAD file according to the different layers
that were identified.
Figure 6 - Segmentation of - the “loggetta” of Sansovina. -
Venice - Italy
The elaboration was then repeated using more rigid research
parameters.
Using this more rigid aggregation method means looking for
the less obvious discontinuity models. This type of
elaboration allows the point cloud to be chosen in smaller and
more homogeneous portions.
This time it was possible to distinguish not only the columns
but also the capitals on different layers.
The presence of many layers on one hand can lead to an
increase in the time that is necessary for the solid modelling,
while on the other it can lead to a better sub-division of the
model and therefore to a more complete model.
