ATING
method to
tral parcels
couples of
| polygons,
mputed by
onventional
intageously
conformal
| points has
lysis of the
id accuracy
lucial point
Procrustes
ated in the
ty of Udine
re national
ed by four
more than
couples of
is reason it
ment new
able direct
cial points,
ed number
nprove the
define the
itum.
solution of
al insertion
liable than
ical map.
ised on the
ith success,
it network,
s cadastral
idjustment-
e cadastral
riangle) of
'anslations,
ain the so-
tisfying, in
> of each
s, the same
update of
the various
iew, a first
astral map,
ion of the
new ones, surveyed in the field with better precision and
reliability. Also in this case, the Generalised Procrustes
algorithm allows the analytical least squares adaptation of the
whole set of the parcels, preserving their original shape.
Finally, given the strict correlation and the geometrical
interdependence existing between the fiducial points and the
vertexes of the surveyed parcels, the proposed procedure of
cartographic updating can be simultaneously applied to the
rigorous adjustment of the fiducial point networks, and to the
optimal integration of the surveyed parcels with those digitised
from the original cadastral maps.
This last procedure of global updating is, from the geometrical
point of view, the most advanced and optimal one, since it
benefits of the constraints and of the high correlation existing
between the geometrical position of the fiducial points, and the
cartographic location of the various parcels.
2. THE RIGOROUS CONFORMAL ADJUSTMENT OF
THE FIDUCIAL POLYGONS
Every time a parcel is surveyed, in order to update or verify the
digital archives, the Italian Cadastre rules prescribe that the
corresponding field measurements must be recorded, processed,
and uploaded into the centralised numerical archives, by way of
a specific computer procedure, freely available, known as the
"Pregeo" program.
As outlined before, the current specifications also state that all
the point coordinates, describing the new shape and position of
the parcels, must be measured with respect to a local reference
frame materialised by at least three fiducial points (fiducial
triangle, or more (fiducial polygon), whose perimeter
completely or almost completely contains the geometric entity
to survey (Figure 1).
Figure 1. Fiducial Point network superimposed to the cadastral
map. Two Fiducial triangles are evidenced
From tasks of this kind it is then possible to extract the
reciprocal distance values existing between pairs of fiducial
points, as determined by the actual field measurements, and to
log them into a file. Univocal and rigorous naming standards are
used to define the fiducial point names at a national level; this
simplifies and makes straightforward every topological
connection required by automated processing.
All this specific information, available in digital form,
represents a precious opportunity to attempt the solution of the
first main cadastral mapping problem listed above, that is the
rigorous re-adjustment of the fiducial point network.
The fundamental philosophy of the proposed approach, is that it
aims at preserving the original shape of every fiducial polygon
(since the term "conformal"). Its geometric configuration in
fact, has been determined by a variable set of angular and
distance measurements, performed among the various fiducial
points and the parcel vertexes, and results from the consequent
internal measurements adjustment. All this furnishes an
"invisible skeleton" to the whole fiducial polygon, giving the
reason of its size and shape. On the contrary, decomposing the
fiducial polygons into their sides, as required by a conventional
least squares adjustment of the distances between fiducial
points, might have the negative effect of destroying these
significant internal geometrical constraints. For these reasons,
the computation of the fiducial point network is carried out
fitting together all the "fiducial polygons" configurations, as
determined by the different surveyors during the time, instead of
processing the single distances.
The procedure that performs these transformations is based on
the Generalised Procrustes Analysis, already employed to solve
a classical photogrammetric problem (Crosilla, Beinat, 2002).
Other recent geodetic applications of the Procrustes analysis are
due to Grafarend and Awange (2000).
Procrustes transformations are widely applied techniques for the
shape analysis, the multifactorial analysis and for the
multidimensional scaling problems (e.g. Borg, Groenen, 1997;
Dryden, Mardia, 1998). They make it possible the least squares
fit of two or more data matrices by searching a proper set of
rotation matrices, scaling factors and translation vectors in order
to satisfy a predefined objective function. The procedure
consents to solve complex multidimensional problems by way
of a rigorous sequence of simple elementary bi-dimensional or
tri-dimensional transformation solutions. One of the major
advantages of the proposed method is that it does not require
any linearization of the equation system involved. As a direct
consequence, the need for approximated values of the unknown
parameters to be estimated is never required.
The full process of conformal adjustment has been structured in
two stages. In the first step, the goal is to establish the most
probable shape, for a given fiducial triangle (or polygon), taking
into account the variability of the different surveys of it carried
out during the time. As a result of the data analysis performed in
this stage it is also possible to compute an index of accuracy for
every fiducial polygon, that can be used as a weight in the
subsequent general adjustment. To this aim, fiducial polygons
of the network, that were never determined directly in the field,
or measured with poorly accurate methods or instruments,
assume a proportionally reduced weight. The second step
concerns the global and simultaneous adjustment of the
representative shapes (consensus) of all the fiducial polygons.
By the same algorithms previously used to define the most
probable shape of every set of homologous fiducial polygons,
the various representative polygons are linked one to the others,
by way of their common points, in order to rigorously
recompose the network of the fiducial points. The procedure
individually rotates, translates, and residually scales all the
polygons is order to obtain their reciprocal best fit condition,
respecting the constraints imposed by an existing control
network (e.g. GPS). This state is achieved when the sum of the
squared residual distances between the homologous fiducial
points, belonging to the various polygon determinations, and
their mean position, satisfies a minimum condition.
Finally, the definitive position of one specific fiducial point is
obtained by averaging the coordinate definitions of the same
point after having rotated, translated, and residually scaled in
optimal way the different fiducial polygons that contain it.
During the adjustment process, the vertexes of the first and
second order of the fundamental cadastral network, and the
vertexes of the National Geodetic network act as fixed points. In
this way not only the adjusted network is constrained to the
control points, letting only local adaptations among the fiducial