International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
Figure 6. Samples of painted
artificial markers
Figure 5. Samples of mobile
artificial markers
These samples helped us in determining the final size of the
markers. Most of the markers have been painted directly on the
ground, mainly on roads or other flat concrete structures. When
this hasn't been possible, in rural parts of the test site, we have
installed artificial mobile ones, made of metal, as Figure 7 and
Figure 8 show.
Figure 8. An installed mo-
bile marker
Figure 7. Laboratory image of
a mobile marker
As previously stated, visibility of the markers is very good for
1:5000 and 1:8000 images and is satisfactory for the smaller
1:18000 scale, as Figure 9, Figure 10 and Figure 11 show.
cr umet
Figure 9. Visibil-
ity on a 1:5000
image
Figure 10. Visibil-
ity on a 1:8000
image
Figure 11. Visibil-
ity on a 1:18000
image
The AGCPs have been measured with GPS in the fast static
mode, using three fixed receivers, set up on vertices of the GPS
network, forming an equilateral triangle. The relative redun-
dancy of the adjustment is therefore three, and results are very
good, as Table 3 shows.
In order to point out and eliminate set-up gross errors, all the
AGCPs have been re-measured.
Min Max Mean Std
e 0.16 0.80 0.437 0.134
n 0.16 0.73 0.321 0.094
0.33 1.50 0.895 0.246
Table 3. Mean statistical parameters of standard deviations s of
the 169 AGCPs, referred to a local Cartesian coordinate sys-
tem, measured in centimetres
4.6 The NGCPs set
Even though the AGCPs play a key role in the Project, we de-
cided to create a smaller set of natural GCPs, because they are
visible in images acquired before the creation of AGCPs: we
have five photogrammetric blocks acquired with a GPS/IMU
aided camera before the start of the Project. Moreover, we are
also interested in estimating the attainable precision on natural
points, which are usually less well defined than artificial ones.
Up to now we have 62 well distributed NGCPs.
Figure 12. Distribution of the NGCPs over the test site
These points have been carefully chosen in order to avoid per-
spective effects, as much as possible, because they must be visi-
ble on many different images. For this reason we looked for fea-
tures belonging to flat surfaces such as roads, courts, etc.
Figure 14. Visibil-
ity on a 1:8000
image
Figure 15. Visibil-
ity on a 1:18000
image
Figure 13. Visibil-
ity on a 1:5000
image
NGCPs have been measured with GPS in the fast static mode,
with the same schema used for AGCPs. The relative redundancy
of the adjustment is three, and results are quite good, as Table 3
shows. NGCPs have been completely re-measured, too, in order
to reject bad data.
Min Max Mean Std
e 0.17 0.76 0.392 0.146
n 0.14 0.79 0.326 0.150
u 0.41 1.49 0.831 0.313
894
Table 4. Mean statistical parameters of standard deviations s of
the 62 NGCPs, referred to a local Cartesian coordinate system,
measured in centimetres
Inte
——
Fou
site
wit]
acq
the
are
heig
1:1
USE:
1:2
cali
ture
are
dep
whi
cali
stru
mat
a cc
IS W
Flig
equ
bloc
Fig
cov
The
vers
bloc
trac
The
The
the
one
hea
end
is 6
arot
The
tute
60/¢
