seconds or minutes of each other, depending on the along-track
angles. The orientation model may be simplified and the number
of orientation parameters reduced, since the satellite orbit is
common to both images. eg. ASTER & OMI.
forward backward
looking looking
image image
—
direction
of flight
Figure 2 - Along-track stereo viewing
2.1. VNIR (Visibl Near Infrared Radiometer) on ASTE
(Advanced Spaceborne Thermal Emission Radiometer)
The Japanese ASTER platform, will be carrying a 3 band
Optical Sensor for Earth Observation (Arai, 1991). VNIR is a
multispectral sensor covering visible and near infrared regions
with a spatial resolution of 15m. The sensor will fly at an
approximate altitude of 705 km imaging the Earth's surface
with a 5,000 element linear CCD (Charged Coupled Device)
sensor. This data is processed on-board, extracting 4,000 pixels
from the full 5,000 pixels of imaged nadir and forward looking
data. The forward looking sensor is set to a viewing angle of
29.7", corresponding to a base to height ratio of approximately
0.6 (Arai, 1991).
2.2. OMI (Optical Mapping Instrument)
The OMI instrument is expected to be flown in the late 1990s at
an orbital altitude of approximately 824 km and inclination of
98.7°. The baseline OMI design is for two views, one 20°
forward off nadir and the other 20° backwards. For a total
coverage of the Earth’s surface in a minimum time, an
across-track capability of £20" will be introduced. The base to
height ratio of the system is approximately 0.7. The sensor is
composed of two 12,000 CCD linear arrays, with a 5m ground
resolution (British Aerospace, 1991).
ERS-1
Resources Satellite-1)
The OPS is an electronic scan typed optical sensor to be flown
on the JERS-1 System . The JERS-1 orbit is a sun synchronous
orbit at 568 km height, with orbital inclination 97.7°. The
sensor has a 4,096 CCD linear array with a 18.3 m range
resolution, and 24.2 m resolution in azimuth. The stereo
capability is obtained by a nadir and a 15.3° forward imaging
sensor, with a base to height ratio of approximately 0.3
(MIT/NASDA, 1990).
3. MODEL DESCRIPTION
Due to its dynamic nature, linear imagery has a distorted
multiple plane perspective, and is unable to be orientated as a
stereomodel obtained from two frame photographs. From
previous studies carried out on SPOT data, it was concluded
that the best models were obtained using the orbital parameters
for the orientation plus the image header file. This approach also
has the advantage of minimizing the ground control data needed.
Consequently, the following orbital method was adopted for the
orientation of along-track stereo imagery.
The geocentiic coordinate system was adopted to avoid
problems of map projection discontinuities, as well as to reduce
the number of formulae, and hence run time. For continuity, the
position of the sensor in space is also described in geocentric
coordinates (Xs,Ys,Zs), which can be computed for each image
line. The image coordinates system (x,y,z) is such that x is the
number of lines in the image along the direction of flight, y is
the number of samples in the across-track direction, and z is the
principal distance of the camera, perpendicular to the image
(Figure 3). The only measurement of time available is x, which
is used to describe time-orientation relationships. However,
since the image is linear, x is assigned to zero, while z always
takes the value -f, where f is the principal distance of the
camera.
P - perigee
S - satellite
Figure 3 - Orbital parameters and geocentric system
The algorithm uses Eulerian parameters ( a, ¢, i, £2, ®, F) to fix
the position of the satellite in space (figure 3) where a is the
semi-major axis, e the eccentricity, i the inclination, Q the
longitude, w the argument of perigee and F is the true anomaly
of the orbit. The elliptical orbit is modelled (using these
parameters) using the rotation matrix Ro (Equations 1 to 4).
R,=Rp Ry Ro (1)
Xs=R, D @)
D=(0,0,r)T 3)
rza(1-e2)/(1-e:cosF) (4)
316
TC
I 5
oi
rc
Ic
Sir
wi
abc
par
dis
set
shc
