Istanbul 2004
polynomial
survey points,
when applied
1 insignificant
Xf control and
'C), supplied
'ormations
good results
software was
iclude RPC,
' are such an
ms represent
eir use is an
nodel of the
ed by means
sis of points,
ts, normally
> used. The
MS error in
respectively.
location of
X points
dy
Pixels
-2.700
-0.598
-1.328
-0.402
0979
-2.342
-0.369
-3.238
-0.820
1.334
-1.521
-2.606
2.388
2.010
-2.387
1.89
3.24
)und
m]
0
6
—
ints
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
After the orthorectification the obtained maximal displacement
and RMS error in position of the check points are 2.30 and 1.54
m, respectively.
The analysis of the results shows that this approach ensures the
needed accuracy for scale 1:5000 in the range of the entire
image.
i WE
B3 01
à 25
& 2;
424 À 20
A16 515
409
D3 22 A 14 4A 23
084
à 26
A 19
106 £17
2s 2
505 A 2i
Ai |
404 |
H 503 |
® Control point à Check point
Figure 2. Points location within the imagery
Since these models do not correspond to the geometric relations
it is better to be applied in cases, when sufficient additional
information is available (points and vector objects), which
information could be used for control of the orthorectified
image.
6. REVISION OF THE TOPOGRAPHIC MAP
The obtained orthoimage is used for revision of the map at
1:5000 scale.
Due to the limitation, imposed by the resolution of the satellite
image, not all objects can be identified with sufficient accuracy
and clearness.
. The boundaries between separate properties are clearly
visible, but hardly classified. The basic information about
them is taken from the available digital model.
. Large residential and industrial buildings of a similar type
are clearly visible on the image and can be plotted in
accordance with the required accuracy.
. The small buildings and buildings with complex outline
are not well recognised. In order to secure their accurate
and proper representation, information from the digital
model of cadastre and additional check (interpretation)
and field surveys, are made.
* The streets, roads and separation strips are easy to be
identified.
. Single train and tram rails are not clearly recognisable.
The stations, however, are clearly visible.
. Water areas — lakes, swimming pools, dams, as well as
their facilities are easily and accurately identifiable.
e. Some of the water flows can be identified indirectly,
while corrected river beds are clearly visible.
e It is impossible to identify the terrain elevation by a single
scene (without a sterco-pair). In order to depict the relief
contours, spot heights and break lines from the cadastre
are used.
When object boundaries are unclear and can not be identified
properly, other sources of information are used:
e Data from the cadastral information.
e Field interpretation of the image.
e Field survey.
7. COMPARATIVE ANALYSIS OF ACCURACY OF
ORTHOPHOTO IMAGES, OBTAINED FROM
AEROPHOTO AND SATELLITE IMAGES
The rapid development of sensor technologies and decrease of
the pixel size on the earth's surface lead to the following
questions:
. Up to what scales and for what purposes the use of acrial
photographs or satellite images is justified?
e Is it justifiable to involve more expensive activities for the
topographic maps at 1:5000 scale?
In respect to this a comparative analysis of accuracy of an
orthophotoimage obtained from satellite image and one from
aerial photographs is performed.
An aerial photography of a part of Sofia with area 4.5 km” is
used. This is a region, having relatively high buildings and
rapidly changing infrastructure. It is covered by 3 strips with 18
aerial photographs in total.
Data about the photography:
° Aerial survey camera RMK A 15/23.
. Approximate image scale 1:4500.
e Average flight height above the terrain 690 m.
The aerial triangulation is based on 8 control points, pre-
marked on the terrain. The achieved accuracy is established by
9 check points. The results are compared in Tables 3 and 4,
where Xm, Y m, Zm are average coordinates calculated from all
models, and Xg, Yg, Zg are geodetically determinated
coordinates, as well as dXY is the ground displacement of the
points. Unacceptable residuals are marked with an asterisk (*).
Ne Xm-Xg | Ym-Yg Zm-Zg dXY
Im] [m] [m] [m]
[1 -0.06 -0.06 -0.26* 0.08
13 -0.01 0.07 0.14 0.07
24 0.02 0.03 0.19 0.04
16 -0.08 -0.07 -0.22* 0.10
31 0.09 0.05 -0.22* 0.11
26 -0.01 0.01 -0.19 0.02
44 0.00 -0.06 0.18 0.06
46 -0.02 -0.07 -0.22* 0.08
RMS: 0.05 0.06 0.20* 0.08
MAX: 0.09 0.07 0.26* 0.11
Table 3. Ground control points residuals
389
