Christian Heipke
4 TEST PROCEDURE
4.1 General workflow
The test consists of two phases (see chapters 4.2 and 4.3 for details). The first phase comprises the determination of so-
called system calibration parameters. The second phase deals with the integration of the GPS/IMU data into the bundle
block adjustment. The object space coordinate system of the whole test is UTM/EUREF 89.
The test participants will receive various sets of test data in both phases and process them using existing experimental or
commercial software. The results will then be communicated back to the pilot centre. They must include the system
calibration data and the exterior orientation parameters of the projection centres of the actual test flight (first phase) and
the adjusted object coordinates of the test field in the second phase. For both phases a report detailing the employed
method of processing, the adopted workflow and a general assessment of the obtained results and problems encountered
needs to be delivered to the pilot centre.
At the pilot centre, the results will be analysed and compared to reference data. Upon preliminary completion of this
analysis a workshop on “Integrated sensor orientation” will be organised. At this occasion the test participants and other
interested individuals can discuss the results and share their experience. In this way a broad dissemination of the results
will be achieved.
4.2 Phase I: System calibration
The first test phase deals with the determination of the GPS shifts, the boresight misalignment and possibly additional
parameters from the information of the calibration flights. In the following the determination of these parameters is
called “system calibration”, which motivates the name of the first phase and also the term “calibration flight”. It should
be noted that the system calibration does not include the camera calibration, i. e. the determination of the parameters of
interior orientation. Only corrections to the calibrated focal length can possibly be computed. Phase I also comprises the
determination of exterior orientation parameters of the actual test flight based on the GPS/IMU data and the results of
system calibration.
The test scheme of Phase I is depicted in figure 4. The camera calibration protocol was provided by the flight compa-
nies. Processing of the raw GPS/IMU data into flight trajectories of the camera projection centre given in UTM/EUREF
89 as well as roll, pitch and yaw of the plane around the camera projection centre as a function of time was performed
by Applanix and IGI, respectively. They also provided the instant of exposure of each image. Interpolation of initial
values of exterior orientation for each image from these data was carried out at IPI. IPI also measured the image coordi-
nates of the GCP and the tie points in the calibration flight images using an analytical plotter.
The first task of the participants consists in the system calibration using a suitable geometric model. There is a choice of
processing the calibration flights 1:5000 and 1:10000 separately, or to use a combined approach. A detailed description
of the employed system calibration model and the computed parameters are the first results of phase I. Subsequently,
the system calibration parameters are to be applied to the GPS/IMU data of the actual 1:5000 test flight in order to cal-
culate the exterior orientation parameters of each image of this flight. These orientation parameters are the second part
of the participants' results of phase I.
The results will be collected by the pilot centre. The analysis of phase I consists of three parts. First, the individual sets
of system calibration parameters and the exterior orientation parameters of the test flight will be compared taking into
account the different system calibration models employed. In the second part, the exterior orientation will also be com-
puted based on traditional aerial triangulation using the image coordinates of tie points and the GCP determined inter-
actively at IPI, and these results will also be compared to the participants’ results. As a third analysis step, the interac-
tively measured image coordinates of the GCP will be transformed into object space via a least-squares forward inter-
section with the exterior orientation of the participants being introduced as constant values. The resulting object space
coordinates will then be compared to the known values of the GCP, and the residuals in image space can be interpreted
as remaining y-parallaxes in stereo models formed using the participants’ exterior orientation.
4.3 Phase II: Integrated bundle block adjustment
The second phase deals with the integration of the GPS/IMU data into the bundle block adjustment in order to obtain an
optimum, i. e. the most accurate solution. The scheme of phase II is depicted in figure 5. After returning the results of
phase I the participants will receive image coordinates of tie points and GCP of the test flight images together with
some additional GCP coordinates in UTM/EUREF 89. Together with the system calibration parameters determined in
phase I they can then perform an integrated bundle block adjustment, estimating the exterior orientation, the system
calibration parameters and the object space coordinates of the tie points and the GCP. These values will subsequently be
returned to the pilot centre together with a detailed report describing the adopted model for the integration. Again, the
pilot centre will compare the results and check them against the known reference values for the object space coordi-
nates, yielding insight into the quality of the different solutions being used by the individual participants.
376 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.
