MODELLING ORIENTATION PARAMETERS OF SENSOR PLATFORMS
Joachim Lindenberger
Institute of Photogrammetry
Stuttgart University
Keplerstr.11 7000 Stuttgart 1
Fed. Rep. of Germany
Commission I / 3
1. Introduction
Recent advances in photogrammetry may be characterized by the
introduction of computer technology into all photogrammetric
procedures. The first link in such procedures, the primary data
acquisition, plays an important role in this development,
because the transformation of a classical photography into a
computer readable format is a time-consuming and costly pro-
cess. For this reason the substitution of the photo-chemical
layer by photo-electrical sensors is the subject of current
research. The three-line scanner and the airborne laser pro-
filing system are two typical examples of the new sensor
technologies which are mentioned here as the background to the
studies presented in this paper.
Comparing modern scanners with conventional photogrammetry the
main principles become obvious : the scanners provide less
information per exposure, which must be compensated by a higher
exposure rate. In consequence the exterior orientation para-
meters of the scanner platform must be determined with a higher
rate. Closely connected to this point is the problem of a weak
geometry of the scanner data. Single spot laser scanning repre-
sents an extreme. When only a single measurement is executed at
every point of time, no geometric redundancy is obtained. Then
the determination of the exterior orientation parameters is not
possible in the conventional, indirect manner with the help of
terrestrial control points. In this case the exterior orienta-
tion parameters must be directly measured by additional devices
with a sufficient accuracy. These brief arguments explain why
the direct measurement of the exterior orientation parameters
of a modern sensor platform and their consideration in the
whole evaluation process is indispensable.
This paper presents a mathematical model describing the dynamic
and stochastical properties of any time-dependent parameter.
The evaluation of measured data takes full advantage of this
model for the solution of a series of problems such as : sto-
chastical description of the measurement process, filtering and
smoothing of observations, detection of gross measurement
errors, stochastical description of the filtered data. To sum-
marize, the dynamic modelling of measured exterior orientation
parameters provides filtered data with their stochastic proper-
ties for input into further evaluations of sensor data.
The measured orientation parameters represent a trajectory of
the sensor platform, which is disturbed by the observation
errors. As the observation errors are unknown, we are unable to
reconstruct the true track of the sensor platform. An approach
to the true movement will be found by the introduction of a
82