In some areas, there are some gross errors 10 the mismatching.
It means that the software has accepted two points as a pair but
in the reality they are not. It is different form the regions that
the software declares them as not matched place. In Figure 12,
some of them are indicated by the red color.
The result of the difference, which is shown in Figure 12, is:
-Median = -14.1m
-Mean -13.5 m
-Standard Deviation = 112.6m
3.2 OrthobasePro v8.7
The OrthobasePro V8.7 has the ability to handle SPOT 5 data.
All the steps and the process are like the OrthobasePro v8.6, but
the differences are the model definition and direct read of
SPOT 5 data format.
3.2.1 Model definition
The first step is to define the kind of model for OrthobasePro
v8.7, which is shown in Figure 13.
Geometnc Model Categon:
| Crbéal Pushbroons |
Camera Panne
Rational Furetions Lancel
Folynomialbased Pushbraor ;
iOrbital Pushbroom i Help
Mixed Sensor
Figure 13. Model definition in OrthobasePro v8.7
OrthobasePro v8.7 uses a model for SPOT 5 called as Orbital
Pushbroom.
3.22 Data Input
The OrthobasePro v8.7 software has the capability to read the
DIMAP (tiff) format, which SPOT IMAGE distributes the data
in that format, directly.
3.2.3 Interior Orientation (Frame Editor)
In this stage, the interior orientation parameters for SPOT 5
could be modified. As the OrthobasePro uses polynomial for
orbit modeling, user should specify the polynomial order for the
orbital parameters such as X, Y, Z, Omega, Phi, Kappa.
3.2.4 Point Measurement
The next step is to select Ground Control Points, Check Points
and Tie Points. For Rasht region, with using the provided
digital maps, the 17 GCPs and CPs (Full Control Point) are
selected. These points are in accuracy of 1:25000 maps. The
selected points are the same as the points selected for PCI
software and OrthobasePro v8.6. Figure 2, Figure 3, and Figure
4 show the distribution of the GCPs and CPs. The digital maps
do not cover the whole imagery area and they cover about 60%-
759^ of the whole images.
Like version 8.6, the ERDAS Imagine also has the ability to
extract the tie points automatically. Because it is tried the
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
conditions to be similar like the other, again 13 automatic tie
points selected. But the tie points are not the same as they are
dependent to the software algorithm to find and extract them.
3.2.5 Triangulation
After doing point measurement process, the bundle adjustment
(triangulation calculation) will be done. To make the condition
the same for OrthobasePro v8.6 and v8.7, the same points are
used. The result of these tests is:
The result of the bundle adjustment for 6 GCPs for each image
and 13 tie points is shown in Table 5.
GCP CP
Ground X (m) 3.967 4.976
Ground Y (m) 4.390 6.513
Ground Z (m) 8.042 10.833
Image x (pixel) 0.001 0.215
Image y (pixel) 0.002 0.004
Table 5. The RMSE of the Orthobase v8.7 bundle adjustment
There are 6 GCPs per image, 5 common CPs, and 13 tie points
Please see the Appendix I for more information about the used
points.
3.2.6 Automatic DTM Extraction
After solving orbit modeling parameters, the software can start
processing for automatic DTM extraction. This software uses
Correlation function for image matching.
3.2.7 The DTM result
Table 6 shows the error analysis on GCPs and CPs in the DTM.
Calculated GCP/CP
ID Elevation (m) Elevation (m) Difference (m)
1 -18.5493 -25.814 7.2647
2 No matching
3 No matching
4 -43.1196 -22.191 -20.9286
5 -6.6655 5.418 -12.0835
6 134.9967 157.167 -22.1703
7 478.7794 499.656 -20.8766
211.0184 218.974 -7.9556
9 14.4012 11.932 2.4692
10 -40.6903 -22.514 -18.1763
11 -34.9393 -22.267 -12.6723
12 -10.0288 1.576 -11.6048
13 45.3006 49.411 -4.1104
14 -25.8908 -22.076 -3.8148
15 -46.4238 -21.47 -24.9538
16 479.3013 474.2 5.1013
17 -8.8323 -11.028 2.1957
548
Table 6. The error analysis on GCPs and CPs in the generated
D'TM by OrthobasePro v8.7
The result of Table 6 could be summarized as:
