International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
+ Since the along track direction of the sensor is the X
direction of the coordinate system and the pixel size in
this direction is 5m, whilst the pixel size in the across
track direction is 10m , it is expected that the accuracy in
this axis should be two times better than the accuracy of
Y axis.
* There are some quite large residuals on some of the
ground control points, and the solution would be
improved if these were removed. However since such
points are characteristic of the control points which can
be determined in such areas, they have been left in.
* The accuracy in height is expected to be better than Sm,
while in mountains areas, this limit is 10m (Valorge,
2003); Airault et al (2003), in a preliminary investigation
of HRS data achieved an standard deviation of 7.4m
From the interpretation of the mean values it seems that
there is a small systematic error in x-direction.
ICP DX (m) DY(m) DZ(m)
Max 15.68 21.67 10.68
Min -10.84 -12.50 -14.15
Mean 0.69 3.30 -0.47
RMS 7.45 9.83 6.41
Table 2. Accuracy of check points from a direct solution in
WGS84
Having these points in mind and also the ability in the direct
solution to make a self calibration process with significant
accuracy, which is not possible when a sensor model is solved
only with GCPs, a modified sensor model is developed in
order to calibrate the focal length of the HRS lenses. The
achieved accuracy is better than 0.085 mm for both lenses and
the calculated focal lengths are 579.86mm for HRSI and
580.36mm for HRS2.
Then the intersection is computed again using the calibrated
values for the focal lengths and the results with respect to the
geodetic system are given in table 3.
ICP Dx(m) Dy(m) Dh(m)
Max 16.82 16.60 8.45
Min -14.44 -15.80 -5.55
Mean 1.83 -1.59 0.20
RMSE 9,66 11.46 5.07
Table 3. Accuracy of the check points after the self-
calibration process.
The accuracy of the check points is improved especially in the
x-.direction and the systematic error is the half of the previous
error but in y-axis. Further work will be done in order to
calibrate and the principal point offset in across track direction.
4.4. Sensor model in the Inertial Coordinate System
4.4.1. Introduction. The model which is developed from the
collinearity equations, which are modified in such a way that
there is no need to use the velocity vector from the metadata
file of each image, based on the assumption that the motion of
the satellite is a Keplerian motion during the acquisition of
along track stereo images. It is the most important of the
models that have already developed because:
The solution of the extended model as it is described in
4.4.2 is the most accurate.
The unknown parameters could be less than the other
models, totally twelve for both HRS images.
The velocity vector information is not used in the solution
giving the opportunity to be used it as independent
variable for checking or as a condition.
4.4.2. Solution of the extended model. In this extended
model the importance of the angular velocity is examined. In
the developed model six parameters which express the angular
velocity vector are added. The total unknown parameters for
the exterior orientation are 18 for both HRS images.
This model could also be solved directly. As it has already
been mentioned, because the rotation angles of the centre
point of HRS images are not calculated correctly, the rotations
are treated as unknown parameters. In this case two GCPs are
needed for the solution where the state vector has already been
extracted from the metadata, correctly. Also and in this
solution a self calibration process is done with significant
accuracy.
Then the intersection is computed again using the calibrated
values for the focal lengths and the results with respect to the
geodetic system are given in table 4.
ICP Dx(m) Dy(m) Dh(m)
Max 18.04 10.12 10.67
Min -15.47 -15.58 -5.74
Mean 1.70 -1.64 -0.16
RMSE 9.78 137 4.68
Table 4. Accuracy of check points from the solution in the
inertial co-ordinate system.
4.5 Discussion.
It has been shown that the SPOTS HRS data can be oriented
using ground control points or metadata and that a solution
can be found which is within the expected error bounds. The
use of self calibration gives a slightly improved solution. We
can conclude that the accuracy when only using orbital data is
good, and that the solution with ground control points is
probably constrained by the accuracy of the control.
5. DEM GENERATION
5.1 Introduction
This section reports on the generation of a DEM using the
beta version of the Leica Photogrammetry Suite (LPS). This
is the upgrade of OrthoBASE Pro which is able to read
SPOTS. The DEM generation is based on area-based
matching which is also called signal based matching. The
LPS is used because the model described above has yet to be
linked to stereo matching software.
5.2 Orbital Model of LPS
In the LPS user manual (Leica 2003) it is described as: “The
Orbital Pushbroom model is a generic model, which is able to
compute the transformation between image pixel (image space,
which is the source image pixel location) and ground point
(ground space, which is the ground point location) for
pushbroom sensors suck as QuickBird, EROS Al, ASTER,
Inter
aS
the
The
pos
In
sol
che
sho
Tab
The
5.31
S.3.]
can |
