Karsten Jacobsen
6 CONCLUSIONS
lax, is The determination of the boresight misalignment includes especially for normal and small angle images the problem of
round a limited separation of the different error components. A flight in opposite direction over a reference area or just strip
n the should be done to enable a sufficient separation of the misalignment in roll and pitch from the constant GPS-shifts in X
s to a and Y-direction. If different flying altitudes should be used, also the reference area should be flown under similar
lative conditions to allow a separation of GPS-shifts in Z-direction from discrepancies of the focal length which can be caused
just by the temperature conditions of the photo flight. For a block flown with a DCS 520 with small angle lenses, a
-0.81, separation of influences from GPS and misalignment was not possible, so the X- and Y-coordinates of the projection
s and centers had to be fixed to the GPS-data for the determination of the reference orientations.
ga is Over the 4 days of photo flight, the change of the misalignment in roll and pitch was within the standard deviations,
only the yaw has changed significant, also during one day. This requires a check of the misalignment every day if the
same photogrammetric camera will be used. It may be different for other camera types.
sensor - ; : ; ; ; ;
; nis is Ground coordinates computed by common intersection based on the direct determined camera orientations of a normal
lax of angle camera and image coordinates in a scale 1 : 3500 have had mean square discrepancies in X and Y in the range of
30cm and for Z 0.95m, sufficient for the project. Problems are existing with the setup of stereo models. The mean
shows n
that square y-parallaxes are 56um, even for 6% of the models more than 100um. The dominating part of the y-parallaxes is
a a constant shift. If this will be removed, the remaining mean square values are reduced to 14um and in no model 30um
t d have been exceeded.
direct Another problem is the missing reliability. Even simple errors of the image numbers can be detected only by a model
‘cason setup or the matching of orthophotos. A combined block adjustment is solving this problems and is avoiding a
disturbing size of y-parallaxes.
ACKNOWLEDGMENTS
Thanks are going to BSF (Berliner Spezialflug Luftbild und Vermessungen GmbH, Diepensee), IGI, Hilchenbach, and
Prof. Dr. Heimes, University of Applied Sciences, Bochum, Germany for the fruitful cooperation.
REFERNENCES
Bäumker, M., Brechtken, R., Heimes, F.-J. (1999): Stabilized and Directly Georeferenced Imagery Based on
INS/D)GPS, Second International Symposium on Operationalization of Remote Sensing, ITC
Enschede, Netherlands
Herrmann, D. (1999): Eignung inertialer MeBsysteme zur Bestimmung von Bildorientierungen, diploma thesis,
University of Hannover, 1999
Jacobsen, K. (1997): Operational Block Adjustment without Control Points, ASPRS Annual Convention 1997, Seattle,
Volumne 2, pp 238 — 244
Jacobsen, K. (1999): Combined Bundle Block Adjustment with Attitude Data, ASPRS Annual Convention 1999,
Portland
Lee, J.O. (1996): Untersuchungen von Verfahren zur kombinierten Aerotriangulation mittels integriertem GPS/INS,
or all doctor thesis, University of Hannover 1996
'eding
hift of
red to
pgram
to be
to be
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 435