International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
b. Preparation of a test-site, execution of test-flights, distribu-
tion of the material to the various units as described in sec-
tion 4 and 5 (Casella, Galetto, 2003).
2.2 Objectives of the second year of research (2004)
c. Definition of refined methodologies based on Kalman fil-
tering of the measurements acquired by the sensors.
d. Refined methodologies for sensor calibration: direct sensor
integration versus integrated sensor orientation (Pinto, For-
lani, 2004).
Stability of calibration of the GPS/IMU sensor.
f. Influence of bad GPS constellation and/or cycle slips.
g. Individuation of suitable methodologies for local parallax
reduction (Casella, Franzini, 2003; Casella, 2004).
h. Individuation of intrinsic methodologies in order to check
the reliability of GPS/IMU data, without resorting to air-
borne triangulation, and without measuring control points
on the ground.
i. Analysis of the main commercial photogrammetry soft-
ware in order to check their functionality related to
GPS/IMU data.
j Drawing up guidelines for the execution and the testing of
GPS/INS aided photogrammetric flights.
3. DEVELOPMENT STATUS OF THE PROJECT
The objectives of the first year have been substantially reached.
Regarding item b., the test site has been completed and meas-
urements are finished. Flights were performed successfully. All
the data has been delivered to the units. The only running task
‘of this part is the measurement of the image coordinates of all
the control points (around 250) and of 18 tie points per image,
for all the acquired images. This is a large job carried out by a
joint effort of four units, that will soon give each researcher the
possibility of using a huge amount of image measurements,
without loading any photogram. Therefore researchers will be
free to perform new measurements, or to focus on the algo-
rithmic part of their research. Moreover, the official image co-
ordinates, after a good data rejection job, will constitute a
common basis for effectively comparing different approaches
and algorithms.
The evaluation of a vast set of images acquired in an industrial
context has been performed, giving interesting results. As this is
a significant part of the Project, the following section describes
its motivation, while the main results obtained are illustrated in
another paper (Casella et al., 2004).
Regarding the activities of the second year, currently (April
2004) all the units started investigating the topics. Some groups
have already published papers related to the items from c. to j.:
the Pavia's unit worked on local parallax reduction (Casella,
Franzini, 2003; Casella, 2004) and Parma's unit worked on re-
fined calibration (Pinto, Forlani, 2004).
3.1 Study of the quality of the exterior orientation obtained
in an industrial context
Quality assessment of direct georeferencing in photogrammetry
is of great interest because this technology is developing rapidly
and the case studies already acquired are significant but not ex-
haustive. Besides, many of the most significant tests, such as
the OEEPE test and those performed by the University of Stutt-
gart, were conducted with dedicated data, which was acquired
892
exclusively for research purposes. Such studies are useful be-
cause they allow estimation of the quality which is attainable
using a certain technology in the best possible way; nevertheless
it is important to ascertain what the attainable quality in an or-
dinary daily work-flow is and how high it's reliability.
This is the precise aim of the described step of our Project,
which has two main goals: a systematic evaluation of residual
parallaxes, which are sometimes critical in directly oriented im-
ages, and a systematic assessment of accuracy. The considered
data is of purely industrial origin and furthermore, the vast and
heterogeneous case study - different cities, different flights, dif-
ferent calibrations - make the described test very significant and
reliable.
To perform this phase the nine working groups, belonging to
different cities, were given a block above each city, together
with EO data measured by an inertial integrated sensor. The
data was supplied by the Italian company Compagnia Generale
Ripreseaeree - CGR, whose headquarters are in Parma, which
joined the Project. The images were acquired before the begin-
ning of our Project, within an industrial programme called Ter-
raltaly-City.
This program has been performed from 2000 to-date and deals
with the production of high detail colour orthophotos for more
than 110 Italian cities at a nominal scale of 1:2000. Images were
acquired with a Leica RC30 camera, provided with an Applanix
POS/AV 510 sensor. The focal length of the lens was 300 mm;
the average relative height of the flight was 2900 m; the average
image scale, 1:9300. The digital images were produced by a
Zeiss SCAI scanner with a resolution of 14 microns per pixel.
The resulting GSD 1s 12 cm.
This part of the Project is of great interest because the nine
blocks considered were acquired before the beginning of our
research Project, following the ordinary daily work-flow of the
company. Results for this part of the Project are reported in
(Casella et al., 2004).
4. PAVIA'S TEST SITE
Pavia's test site has many relevant features which have been
developed in the last five years, according to the needs of the
ongoing research projects. Some further features have been cre-
ated from scratch, to fully support the execution of the Project
under description.
4.1 GPS and levelling network
There is a high-quality GPS network, constituted by 13 vertices.
It was established in summer 1999 and successively extended
and repeatedly re-measured. It has been surveyed with the static
mode and with long-lasting sessions. The network is connected
to the Italian GPS network, called IGM95, which is a materiali-
zation of the ETRS-89 datum. It embraces all of Pavia and its
neighbourhood, as Figure 1 shows. It also includes a GPS per-
manent reference station operated by the Laboratorio di Geo-
matica of the DIET Department of the University of Pavia.
The network is frequently used, for many tasks, so it is continu-
ously under verification. The last adjustment was performed
with a relative redundancy of 8.5 that is, 8.5 baselines were used
to determine each point, in the average. Results are summarized
in Table 1.
Inte
Fi
bet
Tat
th
The
by ¢
of t
heig
geo
aroi
4.2
The
scal
mai
scal
Tab
Mo
pro:
