fully documented in
MS errors referred to
dial distortion. How-
ment have also to be
camera used, ‘flight’
| may result in rather
| photography, image
d below 5°. The pro-
1 rotations about the
eed 15?. Contrary to
ly from 6-tilts about
f Aigosthena exceed-
ly small stereo-bases
ong with liberal con-
tie points determined
e points, which often
ussed here), must be
rly if significant per-
nce of surface relief
imple ground control
in Section 4, regard-
lans as a source for
maging distances are
f image resizing and
era constants or strip
ed).
NG
deling is a key issue
etrically reliable and
ion of very demand-
acies and ‘stretched’
cially available soft-
aces as a DTM with a
ation (more complex
or special treatment;
nmetrically collected
ly integrated by De-
fined by triangles. In
ill the main mode of
rchaeological ortho-
| et al., 2000). Laser
es problems of post-
meshes suitable for
: al., 2001). Besides,
9 laser scanners as it
llection phase to the
nd discontinuities (a
; of experience). But
he type of algorithm
nodel for orthophoto
, 2000). In the expe-
al problem in ortho-
surfaces orthogonal
triangles, a task en-
such cases, the soft-
ction.
ed as a combination
> bottom. This, how-
indom' triangulation
g image resampling.
:quivalent to the im-
ollection scheme has
been adopted. The top of a ‘vertical’ surface patch is described
by a polyline. For each polyline segment (d), three points were
collected at the bottom, as shown in Fig. 1: two corresponding
to its endpoints (A,C) and one (B) approximately to its middle.
This scheme, though somewhat tedious, allows to constrain the
formation of triangles by ‘forcing’ it to follow the surface form
and, thus, secures a possibly faithful modeling (within the scale
tolerance). It is needless to say that only certain ‘difficult’ parts,
not a whole surface, have to be described in this manner. On the
other hand it is clear that good surface description is necessary
but not sufficient: suitable images must also be available as the
result of careful planning.
In Fig. 2 an example is shown from the Aigosthena project. It is
clearly seen that all surface breaks have been faithfully modeled
which helps produce a geometrically correct orthoprojection.
A second example given in Fig. 3 is drawn from the Sparta pro-
ject. There, a view of the western parodos is shown, along with
Figure 1. Breakline and points forming ‘vertical’ triangles.
Figure 2. Shaded surface model of a part of the Aigosthena Castle (top). Below are also are seen a detail of the surface model by the
window, showing the ‘vertical’ hang, and the corresponding area of the orthomosaic.
the shaded model of a detail area, whose orthoimage is also to
be seen. The full orthomosaics of the two parodoi are presented
in Fig. 4, whereas in Fig. 5 a further example of surface model-
ing from the eastern parodos is given. Finally, the products of
the Zea project are illustrated in Fig. 6.
4. USE OF EXISTING PLANS AS CONTROL
As already mentioned, an aspect regarding cost efficiency in the
production of digital orthomosaics concerns the exploitation of
pre-existing line drawings, plans or elevations of a site. Indeed,
E
