and nght) and with image matching techniques and
taking correlation between conjugate pixels, we can
create a DEM. However, in this project since our aim
was not creating DEM, but using it for different
purposes, we used an existing DEM of the area that had
already been created from digitizing contour lines in
map and it was georeferenced and covered all our area
of interest and thus we used this existing DEM .
3.3 Extraction of topographic data from
scanned aerial photos
3.3.1 Scanning photos and format conversion. Aerial
photos were scanned, to convert analog data into digital
raster form. The required scanning resolution was
chosen to suit both accuracy and storage capacity. Thus,
the resolution of 100 dpi was appropriate. The raster
data was in the TIFF format, therefore, in order to be
used in the ILWIS system, it should get converted to
MPD and MPI formats. The Data Conversion program
was used for this process.
3.3.2 Radiometric correction . Sun angle correction is
not applicable since we had only one channel
information per scene. In this particular case, therefore
only linear stretch was applied. For haze correction, we
just shifted the brightness histogram of scanned photos
to Zero.
3.3.3 Creation of orthophoto. From the beginning, our
area of interest was well defined in terms of both X- and
Y- coordinates so only two scanned photos covered the
whole area. therefore, the 2 orthophotos had to be
created. For this matter, the planimetric coordinates
were interpolated from existing map.
3.3.3.1 Selection of Ground Control Points (GCP).
For a good georeferencing, GCPs need to be well
distributed all over the image. For that matter, tie points
are selected in the overlapping area. The 1:25000 map
was used to acquire the coordinates of the well-defined
points on both the map and photographs. The list of
GCPs is shown in figure 3 for photo 36 and 37.
3.3.3.2 Orientation of scanned photos. ORIENT
program performs an inner orientation based on the
given relation of the fiducial marks provided the
following data are given:
1. Control point file
2. Patched DEM file
3. Calibrated focal length
The transformation file will be delivered as output.
3.3.3.3 Formation of orthophoto. Once the orientation
is satisfactory, the patched DEM, the patched scanned
image, the output orientation files are required by
ORTHOPHoto program. The output image is still patch
form. Thus it has to be unpatched which will deliver
already to display digital orthophotos.
538
T36N.DAT iu
pn Col Line x y use nameesg
1 845 461 0.115 0 1 east.
2 430 876 0 -0.115 1 south
3 20 457 -0.115 0 1 west
4 436 42 0 0.115 1 north
-1
pn x X Y use
1 767.0 878.0 840338.0 3173500.0 1
2 300.0 794.0 838525.0 3173638.0 1
3 45.0 599.0 837450.0 3174500.0 1
4 20.0 94.0 837250.0 3176575.0 1
5 499.0 39.0 839300.0 3176863.0 1
6 759,0 183.0 840375.0 3176250.0 1
7 828.0 416.0 840650.0 3175288.0 1
T37NN.DAT
pn Col Line X y use namees$
1 845 449 0.115 0 1 east
2 428 863 0 -0.115 1 south
3 19 443 -0.115 0 1 west.
4 436 30 0 0.115 1 north
~1
pn X X Y use
1 667.0 771.0 838525.0 3173638.0 1
2 395.0 575.0 837450.0 3174500.0 1
3 217.0 792.0 836713.0 3173625.0 1
4 189.0 497.0 836550.0 3174800.0 1
5 360.0 74.0 837250.0 3176575.0 1
6 725.0 185.0 838813.0 3176150.0 1
figure 3- GCP file
3.34 Clipping and Merging. After creation of
orthophoto, both orthophoto 36 and orthophoto 37 were
merged. The beginning of column of each orthophoto
was removed to avoid some darkness after mosaicking.
With the program MERGE, both orthophotos were
merged together and meanwhile we defined an area of
interest from the map and scanned photographs.
3.3.5 On screen digitizing and Editing. The displayed
raster image on the screen, without using any hard copy,
identified features were being digitized directly in the
forn of point and vector. With the advantage of
zooming in and out, the cursor movements are adapted
to the whole screen and the cursor becomes sensitive
because of following the coordinates of the map placed
on the digitizer. Several possibilities have been
introduced with respect to the digitizer set up and the
digitizing colors and even the existing codes are
displayed when the user decides to change the mask and
the codes of the segments. Thereafter, all the digitized
data in the segment mode, are polygonized in order to
make closed polygons. For this process, a unique
number is assigned to each polygon according to the
landuse separation.
4. IMPROVED CLASSIFICATION
4.1 Geometric correction
The following steps have be done:
- Patching the radiometrically corrected satellite images
by use of PATCH program, since RAM does not have
sufficient space to handle the whole image:
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996