ul 2004
we de-
hey are
IPs: we
"S/IMU
we are
natural
il ones.
ite
jid per-
be visi-
for fea-
sibil-
000
; mode,
ndancy
Table 3
n order
)
)
)
ns s of
ystem,
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
: Focal Relative Strips Images
Flight Date Scale Date flight Overlapping SIPs mages
length heioht number number
t=)
1:5000 14/05/03 150 mm 750 m 60/30 8 139
1 14/5/2003 1:8000 14/05/03 150 mm 1200 m 60/60 B 131
1:18000 14/05/03 150 mm 2700 m 60/60 2 19
1:5000 16/05/03 150 mm 750m 60/30 135
2 16/05/03 1:8000 16/05/03 150 mm 1200 m 60/60 11 128
1:18000 16/05/03 150 mm 2700 m 60/60 2 15
3 06/0403 1:5000 06/04/03 300 mm 1500 m 60/30 8 146
1:8000 06/04/03 300 mm 2400 m 60/60 11 145
4 17/03/03 1:8000 17/03/03 300 mm 2400 m 60/60 11 135
Table 5. Summary of the performed flights
5. FLIGHTS PERFORMED OVER PAVIA'S TEST SITE
WITHIN THE FRAME OF THE PROJECT
Four different flights have already been performed over the test
site, by the Italian company CGR, whose planes are equipped
with Applanix POS/AV 510 sensors. Two of them have been
acquired with a camera whose focal length is 300 mm, while
the others have been taken with a 150 mm camera. The flights
are composed of a certain number of blocks, flown at different
heights and characterized by the scales 1:5000, 1:8000 and
1:18000. As previously stated, these image scales are usually
used in Italy to produce maps respectively at the scales 1:1000,
1:2000 and 1:10000. Flights are usually distinguished between
calibration and test flights, in the direct photogrammetry litera-
ture. The first are used to calibrate the sensor, while the second
are used to assess precision and quality. They should be as in-
dependent as possible. We chose to perform complex flights
which allow the user to follow several different strategies of
calibration and testing. Moreover, the blocks have a complex
structure themselves, to fulfil the need of independently esti-
mating the calibration parameters. Another motivation for such
a complex structure is to allow intrinsic quality assessment, that
is without external control measurements.
Flights 1 and 2 have been acquired with a Wild RC30 camera,
equipped with a 150 mm lens. They are composed of three
blocks whose structure is shown in Figure 16, Figure 17 and
Figure 18. The 1:5000 block has three ordinary parallel strips
covering a part of the test site, flown in an East-West direction.
The first strip, once completed, is immediately re-flown in re-
verse. There are two cross strips, at the head and tail of the
block; each of them is re-flown in reverse at the end. The along-
track overlapping is 6096, while the across-track one is 309^.
The number of images taken is around 140.
The 1:8000 block has seven ordinary parallel strips covering
the whole test site, flown in the East-West direction. The first
one is flown back and forth. There are two cross strips, at the
head and tail of the block; each of them is flown back (at the
end) and forth (at the beginning). The along-track overlapping
is 60%, as well as the across-track. The number of images is
around 130.
The 1:18000 block has a very simple structure and is consti-
tuted by two strips flown in the East-West direction, with the
60/60 overlapping. The number of images taken is around 20.
Figure 17. Struc-
ture of the 1:8000
block
Figure 18. Struc-
ture of the 1:18000
block
Figure 16. Struc-
ture of the 1:5000
block
Flights 3 and 4 have been acquired with a Wild RC30 camera,
equipped with a 300 mm lens. Their structure is similar to that
of flights | and 2, but not the same. Indeed, we decided not to
acquire images at the 1:18000 scale, because this would have
required a high-altitude flight, and good results were not guar-
anteed. Flight 3 is composed of the 1:5000 and 1:8000 blocks,
having the structure described above. Flight 4 is composed only
of the 1:8000 block.
Table 5 summarizes the main parameters of the flights. It is no-
ticeable that the total number of images is around 1000. Consid-
ering they are colour images, scanned with a pixel size of 14
microns, each image occupies 800 MB. The whole set has a size
of 800 GB, corresponding to 200 DVDs. As we use tiled TIFF
files, JPEG-compressed, the size of each image can be reduced
‘to 180 MB and the whole set corresponds to approximately 50
895
DVDs.
It is clearly possible to conceive various strategies for calibrat-
ing and testing the flights. It is also possible to calibrate with
only one block or with two or even three, in order to be able to
estimate the true focal length. As we have already pointed out,
the complex structure of the flights, has the main advantage of
allowing the uncorrelated estimation of the calibration parame-
ters and of the camera self-calibration model.
6. CONCLUSIONS
The paper describes the research Project, the test site and the
test flights. The Project is under development: although it is
well defined, it is still open to adjustment, depending on the re-
sults of the Workshop. All the scheduled flights have been per-
formed, although it is possible that we will decide to undertake
further flights, in case it should prove useful.
