AUTOMATED GENERATION OF COLOURED ORTHOIMAGES AND IMAGE MOSAICS
USING HRSC AND WAOSS IMAGE DATA OF THE MARS96 MISSION
Frank Scholten
Technical University of Berlin, Department for Photogrammetry and Cartography
Sekr. EB9, Straße d. 17. Juni 135, D-10623 Berlin, Germany
Phone: +49-30-314 23991, Fax: +49-30-314 21104, E-Mail: franky Q fpk.tu-berlin.de
Commission Ill, Intercommission Working Group II/III
KEY WORDS: Automation, Geometry, Radiometry, Orthoimage, Mosaic, Multispectral, Extraterrestrial
ABSTRACT:
The Russian Mars96 Mission is going to be launched in autumn 1996. It will carry the German cameras HRSC (High Reso-
lution Stereo Camera) and WAOSS (Wide Angle Optoelectronic Stereo Scanner). This combined experiment will provide
multiple along-track stereo imagery of the whole planet Mars. Orthoimages and coloured image mosaics will be generated
in completely automated processes. Orthoimage generation uses geometric calibration information of the CCD line scan-
ners HRSC and WAOSS, orbit and pointing information for each line of the entire strip of image data (improved by photo-
grammetric bundle adjustment) and Digital Terrain Models (DTMs) derived by multi-image matching and special inter-
polation techniques. While the geometrical aspect of mosaicking will be taken care of within the bundle block adjustment
radiometrical mosaicking has been improved in that way, that all definitions which had to be made in advance, e.g.
definition of overlapping regions and division lines, are now performed automatically and not interactively on the screen
of an image processing system. High quality coloured mosaics can be generated afterwards using IHS colour trans-
formation in order to preserve high resolution as well as colour information. Especially true colour Topographic Image
Maps in scales 1:500,000 up to 1:50,000, containing contour lines derived from a DTM, are the desired products of this
cartography-oriented mission to planet Mars.
1. GENERAL ASPECTS
OF HRSC AND WAOSS IMAGE DATA
Figure 1:
The HRSC and WAOSS cameras (Albertz et al., 1992) are a) left: Martian surface
very special line scanners the commanding strategies of
which allow different types of imaging sequences of
inhomogeneous structure. Thus the beginning of image
lines may start at varying sample positions of the imaging
CCD array within one data set, may have a varying
number of samples per line, and even gaps between
image lines may appear due to possible loss of data
during transmission or decompression. Furthermore the
scale of a pixel will not be constant within one data set
because of the extreme elliptical orbit of the spacecraft.
The scale factor may vary within one imaging sequence
by the factor of approximately 10. Other commanding
strategies will also form so-called macropixels by combi-
ning 2 - 2 or up to 31 - 31 pixels to one macropixel in order
to generate a nearly constant resolution on ground. Last
but not least the commanded time intervals between the
lines will not be constant due the elliptical orbit.
b) lower left: simulation for
an elliptical orbit and
constant macropixel format
c) lower right: simulation for
an elliptical orbit with
varying macropixel format
These powerful commanding strategies offer an enormous
variability of imaging sequences in order to optimize data With respect to mapping purposes HRSC and WAOSS on
acquisition to the orbit geometry and the scientific re- such an elliptical orbit will provide imagery for large scale
quirements. On the other hand traditional image proces- topographic image maps in scales up to 1:50,000 as well
sing techniques have to be adapted and improved in order as global views of the planet from limb to limb, appropriate
to enable operational and fully automated processing. for smaller map scales.
Fig. 1 shows part of the planet's surface as it is (a), This is also a challenge for automated mapping software
imaged from an elliptical orbit with imaging distance vary- because of the great variety of different map projections
ing by a factor of 2 (b), and imaged from the same orbit with their sometimes critical behaviour in such extreme
with macropixel formats varying from 5 to 10 (c). geometric constellations.
351
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996
