Michael Breuer
of ground control points. Globally applied polynomial functions normally do not meet the requirements because of the
local character of the image distortions. Hence a “Continuous Piecewise Method” is proposed by Ji and Jensen (2000).
Another approach is the “Facet Method” described by Buiten (1993). This method is also referred to as “Rubber
Sheeting”. Here a triangulation is performed over all control points. Then a polynomial interpolation is performed
inside each triangle. This method is suitable to correct geometrically the local deformations in the image. But problems
occur outside the mesh with extrapolation.
6.2 Parametric approaches
Within a parametric approach all parameters of the time-variant image capturing process can be modeled rigorously.
Normally this is done within an adjustment procedure, which allows the modeling of error propagation. With a
parametric approach it is possible to model the inherent disturbing effects individually because of the fact that even
correlation between observations can be modeled if they are known. However, these methods need adequate initial
auxiliary information like position and attitude data, a DEM and/or a reference orthoimage. Pope and Scarpace (2000)
propose a method based on "Image-to-Image-Matching". This method makes use of a reference orthoimage and a
DEM. The aim is to reconstruct a virtual flight path that can be used for subsequent direct georeferencing. Initial
position and attitude data serve as approximations that are improved later on. Schläpfer et al. (1998) also try to
reconstruct the flight path but based on ground control and initial position and attitude data.
6.3 Mixed approaches
A mixed approach contains elements from both, the parametric and the non-parametric methods. These methods are
often applied in cases where the initial information for position and attitude are available with low accuracy only. The
method that was described by Breuer and Albertz (1996) is a mixed approach. In a first step a transformation of the
flight path is calculated using the position and attitude data in combination with ground control. This is done using a
parametric model. In a second step a polynomial transformation is applied to wrap the image data to the terrain taking
into account a DEM and ground control points.
7 CONCLUSIONS
High quality geocorrection is crucially important for everyone who wants to use hyperspectral data. On the other hand
auxiliary data are not always available in an appropriate manner. This reality must be recognized. Therefore up to now
different methods were developed to solve the problem on the background of very special initial circumstances. But the
paper states that a special adoption due to an individual sensor system is needed for the first processing step only (see
5.1). The subsequent steps should be treated holistically taking into account all auxiliary information that can be
available (see Fig. 2). Following this principle an optimal accuracy level should be reachable if the appropriate
algorithm is used in an individual case (see Fig. 3). The future development has to be concentrated to put together the
existing models to create a modular system that can respond to the need of the user who wants to get an optimized
solution based on the real data that he has. At the same time future investigation is needed to work out the accuracy
level concept in more detail especially regarding the quantitative statements.
REFERENCES
Albertz, J., 1998. The Geometric Restitution of Line Scanner Imagery — Three Decades of Technical Development.
Wissenschaftliche Arbeiten der Fachrichtung Vermessungswesen der Universität Hannover, Nr. 227, pp. 25-34.
Albertz, J., W. Zhang, Z. Li, R. Uebbing, G. Wu, 1995. Geometrische Korrektur von Daten flugzeuggetragener
Abtastsysteme unter Einbeziehung von Fluglageparametern. Scientific Report of the BMFT project 01 LF 9202/5, TU
Berlin, 80 pages.
Avery, T.E. and G.L. Berlin, 1992. Fundamentals of Remote Sensing and Airphoto Interpretation. Fifth Edition. Max-
well Macmillan Canada Inc., Don Mills, Ontario, pp. 112-119.
Binnenkade, P., 1993. Multispectral Scanning. In: Buiten, H. G. and J.G.P.W. Clevers (Eds.), Land Oberservation by
Remote Sensing — Theory and Applications, Gordon and Breach Science Publishers, Reading, pp. 137-153.
Breuer, M. and J. Albertz, 1996. Geometric Correction of Airborne Line-Scanner Imagery. IAPRS, Vol. 31, Part B3,
Commission III, Vienna, pp. 19-23.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 99