International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
A significant portion of the total cost of producing digital
orthoimages is that creating the Digital Elevation Model
(DEM) and ground control points (GCP) required in the
orthorectification process. Therefore, it seems logical to
attempt to realize cost reductions from this part of the
production workflow. If DEMs generated from various sources
can be used to orthorectify medium spatial resolution image (5
to 100 m. spatial resolution) such as SPOT imagery and the
number of GCPs is decreased, then the orthoimage production
cost can be reduced significantly (Erdogan, 2000).
Various studies were performed about orthoimage generation
and their accuracy assessments. Different methods and data
inputs were tested to achieve better accuracy. In one of these
studies conducted by Chen and Lee (1993), an
orthorectification method for SPOT images developed by the
authors was tested and the accuracies of the produced
orthoimages were assessed. They found that accuracies better
than two-thirds of a pixel could be achieved with their
orthorectification methods.
Another study conducted by El-Manadili and Novak (1996)
was again the orthorectification process. They used a Direct
Linear Transformation Model for precision rectification of
SPOT Imagery. Especially, they worked on the effects of the
number and quality of ground control points and base-to-height
ratio of the images over the accuracy of produced orthoimage.
The results show that sub-pixel accuracy can be achieved
similar with the above study.
Heipke et al. (1992) tested SPOT Imagery for point
determination, DEM generation and orthorectification with the
automatic photogrammetric processing.
In a study conducted by Radhadevi et al. (1994), the geometric
correction accuracy of SPOT stereo pairs was tested by using a
single ground control point for orbit attitude modelling. Terrain
coordinates are derived up to an accuracy of 28 meters in
latitude and 40 meters in longitude and 27 meters in height
with only 1 ground control point.
Pala and Pons (1995) tested the incorporation of relief in
polynomial-based geometric correction of SPOT and Landsat
TM imagery. It is investigated that sub-pixel accuracy can be
reached with this method.
Above studies show that the accuracy orthorectification process
and the orthoimages are the concern of remote sensing society
yet and many researches have been performed to develop the
accuracies of these products, to reduce the production time and
cost. Since SPOT imagery is widely used by public for DEM
and orthoimage production, SPOT imagery has been used in
many of these researches. When the needed number of ground
control points is decreased significantly in that process,
objectives are accomplished.
2. SPOT SYSTEM AND THE SAME PASS
CONSTRAINTS
2.1 SPOT System
SPOT satellites began a new era in space remote sensing, as it
is the first satellite system to include a linear array sensor and
employs pushbroom-scanning techniques. It is also the first
system employing pointable optics. This enables side-to-side
off-nadir viewing capabilities and it efforts full-scene
stereoscopic imaging from two different tracks allowing
coverage of the same area.
The SPOT sensors can acquire stereoscopic image pairs for a
given geographic area. This is achieved by making two
observations on successive days such that the two images are
collected at angles on either side of the vertical. Stereoscopic
imaging capability of SPOT allows generating DEMs from a
pair of overlapping images. DEMs based on satellite images
are essential for many applications when you need up-to-date
and cost-effective information about terrain relief. Topographic
mapping contouring and orthoimage generation are the two
widely used application areas. A study conducted by
Theodossiou and Dowman (1990) has shown that SPOT data
could be used for mapping at 1:50.000 scale with 20-m.
contours. And that if the data are very good and the ground
control is sufficient, 1:25.000 scale plotting may be possible.
Toutin and Beaudoin (1995) applied photogrammetric
techniques to SPOT data and produced maps with planimetric
accuracy of 12 m. with 90 percent confidence for well
identifiable features and an elevation accuracy of 30 m. with
90 percent confidence.
2.2 Same Pass Constraints
The accuracy of orthoimages is a function of many variables,
one of which is the accuracy of the ground control information
used in a simultaneous adjustment and updating the satellite
model parameters. The satellite model used in orthorectifying
the images is a mathematical representation of the physical law
of the transformation between the image and ground spaces. It
corrects the entire image globally and also takes into
consideration the distortions due to terrain. Unlike the
polynomial models, which require a large number of well
distributed GCPs in order to avoid degradation of the model in
some part of the image, the required number of GCPs is lower
in the mathematical model. This is important when the control
points are acquired by expensive differential GPS
measurements.
624
GCPs are used to calculate the position and orientation (i.e.
roll, tilt, and yaw) of the imaging system at the moment of
image taken. This calculation is accomplished using standard
photogrammetric algorithms, such as a space resection or
bundle adjustment. The position and orientation of the imaging
system are expressed as six values: x, y, z, roll, tilt, and yaw
(or alternatively, x, y, z, omega, phi, and kappa), which
collectively define the exterior orientation of the imaging
system for each image. They are needed in order to map each
pixel of the digital image to its precise location on the ground.
These processes can change according to the characteristics of
the imaging system.
The other issue to consider include the distribution of ground
control in the image and the requirement for additional control
points to provide redundancy. Ground control necessary for
producing orthoimages often comes from ground surveying
which would help to reduce the propagation of the error source
into the orthophoto pixel positions. However, ground surveying
techniques are usually costly. Alternatively, the ground control
measurements may come from the aerial photographs or from
hardcopy maps of the project area.
Orthoimage generation uses the method of space resection to
determine the relation between the object space and image
space. Thus, the accuracy of the ground control used in the
Inter
proc:
absol
quali
Some
deliv
of d
satell
the c
offer:
In thi
adjust
invest
neede
better
which
used.
groun
homo;
the Fi
the m
block
results
