International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
2. GROUND CONTROL POINTS
Ground control points were established in order to produce
1/5000 scale standard maps by photogrammetric mapping
method. All control points are signalized and measured by
GPS. In the project area there are14 signalized ground control
points. After adjustment of GPS measurements accuracy of
ground control points less then 10 cm. All control points in
test area re-signalized for the satellite imaging. Signalization
dimensions are shown below figure. After signalization of
that control points satellite images were provided from INTA
Company.
© © © ©
© ©
© ©
© © ©
© © ©
Figure 1- Ground Control Points
[ On
| m ri
Figure 2- Signalization of GCP's
3. SATELLITE IMAGES SPECIFICATIONS
A pair of stereo images provided free of charge by INTA
Space Company in order to use test works. Image overlaps is
9,100 and time interval between two image around |
minutes. Ground resolution of first image is on cross track
0.88 m and on track 0.85 m. Scanning azimuth 179.96
degrees, nominal data collection azimuth 78.6370 degrees,
nominal data collection elevation is 74.57693 degrees.
Azimuth of sun is 134.9709 degrees and sun elevation
68.83922 degrees during scanning. Ground resolution of
second image is on cross track 0.92 m and on track 1.02 m.
Scanning azimuth 179.96 degrees, nominal data collection
azimuth 163.4890 degrees, nominal data collection elevation
is 62.21244 degrees. Azimuth of sun is 135.4087 degrees
and sun elevation 68.95507 degrees during scanning.
Pan Sharpened stereo images covered whole test area.
Geometric resolution is 1 meter and spectral resolution is 4 m
of those images. More information about Ikonos images
please visit; http://www .spaceturk.com.tr
1164
4. TEST PHASES
4.1 ACCURACY INVESTIGATION
Inside the test area, two different sub areas have been chosen
for accuracy investigation. One of the regions has very dense
parcels, flat area, definition and identification of boundaries,
details would be easier and will be called 4B area. Second
region has mountainous, forestry related parcels, definition
and identification of boundaries, details would be harder and
will be called 9B area.
4.1.1 Preliminary Investigation Of Images
SOCET SET V.4.4.2, ERDAS IMAGINE V8.5, V8.6 and
PCI GEOMATIC V7.2 software were used for different
purposes during test phase.
Before starting test phase, satellite images investigated
detailed and realized following findings;
> Size of each part of Pan-Sharpened stereo images with
1 meter ground resolution is around 400 MB,
» There is a specific file called RPC contains sensor
parameters, meta data and other information,
Images covered whole test area,
» All signalized ground control points are identified and
defined,
Some part of test area covered by clouds and there is no
information about terrain on the cloudy area,
» Some milky areas has a little terrain information which
can interpretable, 4
» There is sun light reflection on the building roof surface
on settlement area., so it is not possible to collect data
about these buildings,
> Linear features such as parcel boundaries can
interpretable but attributes of linear features can not be
defined such as fence, hedge and so on,
> Some of linear features presents is around 3-4 pixels. It
seems a gap between linear features. So, It needs a
special experience to identity if is it patches or drainages
and so on,
> Some details for example fountain, communication line
pillar, power line pillar in the settlement area can not be
seen on the satellite images,
> Main power line and its pillar can be seen easily and
identified very well,
> In very dense agricultural area parcel boundaries can be
identified exactly,
» [n mountainous and forestry related area, definition and
identification of parcel boundaries and its details can be
very hard.
/ YY
v
4.1.20rientation Of Images By GCP’ s
There are 14 ground control points in the test area. Ground
control points datum is European Datum 1950 (ED50). All
control points measured on images. We used DLT (Direct
Linear Transformation) for image orientation. Orientation
results are shown below;