=
room
0 CO-
rman
loped
go as
enta-
scale
par-
>xter-
Con-
kage
been
way.
ictice
| cer-
| per-
imag-
cribed
ssions
mina-
S, this
jvesti-
tation
) and
others treated the problem from the general photogrammetric
triangulation point of view by extending to several longer
strips and using all three-line images. We have restricted our
work to single stereo scenes formed by the two fore and aft
stereo channels. As the spatial movement of the camera over
relatively short segments may be assumed reasonably deter-
ministic and stable, orbit and attitude has been modelled by
simply describing the six orientation parameters as second or
third degree polynomials in time. The functional approach to
kinematic interior and exterior orientation and the theory be-
hind the photogrammetric adjustment and stochastic model of
the utilized constrained collinearity equations is described in
detail in (Dorrer et al., 1995a).
We developed a program package, denoted ORIMOMS, writ-
ten in C, which enables the determination of kinematic exter-
ior orientation of a MOMS-02 stereo scene from measured
image coordinates of a sufficiently large number of tie and
ground control points (Dorrer et al., 1995a). ORIMOMS is
imported in Zeiss' Planicomp analytical plotter data base man-
agement software PHOCUS and may thus be called within
MOMS (see next section), but can also be utilized as stand-
alone version. It identifies the type of spheroidal map coordi-
nates (e.g. UTM) of the given ground control and performs a
transformation to a topocentric cartesian system local to the
current stereo scene. On a SGI R4000, processing time for
ORIMOMS with some 100 model points requires less than 30
seconds. Experience so far shows that in order to minimize
model parallax, tie points must be selected in insufficiently
controlled areas. This is mainly caused by high correlation
between some of the orientation parameters and could be rela-
xed by employing a different functional model.
INTEGRATION IN PLANICOMP-PHOCUS
Analytical stereorestitution of MOMS-02 hardcopy images
with analytical plotters is supposed to function under the same
basic software system used for the compilation of "normal"
metric frame photogrammetry underlying perspective geome-
try. A software package, denoted MOMS, has been developed
enabling precision 3D-restitution of MOMS-02 stereo scenes
on the Planicomp P-series analytical plotters of Zeiss. Basis is
a real-time program module running in background mode un-
der VMS on VAXstation or UNIX on Silicon Graphics Indigo
as host. Considering the typical kinematic imaging geometry
for linear array sensors, MOMS has been rigorously integrated
in the latest version of the photogrammetric restitution soft-
Ware system PHOCUS (Fig. 2).
VAXVWMS . SGI/UNIX
i Planicomp Driver :
Figure 2: Experimental MOMS software integration into
PHOCUS environment.
541
Several necessary or useful interactive functions are available,
e.g. determination of interior orientation; determination of
exterior (absolute) orientation via the program ORIMOMS,
unless already known from external sources; semi-automated
mono- or stereoscopic measurement of image points; 3D-
measurement of model points and contours; transformation of
object coordinates between different mapping systems
(presently UTM, Gauss-Krüger, Spheroidal, Geocentric, To-
pocentric). The influence of earth curvature on elevation con-
tour lines as well as geometric nonlinearities of the individual
CCD-lines (Dorrer, et.al., 1995c) may be considered by real-
time corrections within MOMS.
Since the size of analog pictures on the Planicomp stages is
limited to 230 by 230 mm, particular emphasis must be placed
on a simple way of extracting specifically defined stereo
measurement scenes from the original image data format. For
this reason, a separate, interactive program package,
MOMS GIID, running under UNIX and VMS on workstations
equipped with X11-libraries, has been developed. It contains
several useful functions permitting direct on-line access to line
and frame header information of the image data, e.g. image
roaming and zooming, simple image enhancement, specifica-
tion of individual stereo scene formats, image data output for
the production of hardcopies on a precision filmwriter, etc.
For a detailled description of the main features of the entire
software package see (Dorrer et al., 1995b)
ACCURACY POTENTIAL OF 3D-MEASUREMENT
In essence, two larger experiments were performed with data
taken over the Andes in Bolivia (orbit 115) and the Out-Backs
in Australia (orbit 75b). AII image and stereomodel measure-
ments were carried out in a Planicomp P2 analytical plotter on
diapositives with 40 mikron pixel size yielding 1:337,500
image scales for channel 6/7 stereoscenes (ground resolution
13.5 m). The diapositives were produced on a Cirrus LC-
3000 Laser filmwriter at GAF Image Processing Services.
While ground control for the Bolivia scene could only be deri-
ved from an old 1:50,000 topographic map, the Australia
scene was fully controlled by over 50 GPS-determined ground
points with sub-decimeter accuracy (Fraser et al., 1996).
Therefore, trustworthy accuracy assessment results, to be
discussed in the sequel, could only be expected from the Aus-
tralia scene.
The Australia scene covers a fairly flat and featureless terrain
in the Lake Nash/Georgina River area of the southeast Nor-
thern Territory. Maximum elevation differences are below
80m. Even though ground control was referred to man-made
features, e.g. corners of small dam sites, intersections of
country roads or fence lines, centers of water holes, point
identification in the imagery turned out to be nontrivial. See
also (Fraser et al., 1996) for details. Stereoscopic interpreta-
tion and identification proved extremely helpful, and must be
strongly recommended for future work. In the course of de-
termining the kinematic exterior orientation with ORIMOMS, 7
control points had to be rejected for obvious gross errors.
Some of these could be corrected due to misinterpretation and
were reconsidered. Together with 41 measured tie points, the
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996