Ш
Ш
чШ
ш
.
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008
Layer
No
Feature Names in Layer
Number of
Detected
Features
Compilation
Percentages
Aerial
Photo
Quickbird
£
5
Rocks and stony place
22
-
0.0
31
Water depot
1
-
0.0
42
Telephone/Radio line/station
2
-
0.0
43
Sporting facilities
6
-
0.0
44
Single grave or graveyard
2
-
0.0
49
Pipe line and sewerage
3
-
0.0
51
Ditch, set and tumulus
43
-
0.0
27
Lean-to roof
397
13
3.3
48
Telephone/Electric pole, lamp
321
32
10.0
20
Pavement
183
30
16.4
50
Slope and natural split
183
31
16.9
12
Water well and canal
9
2
22.2
45
Electric line and transformer
53
18
34.0
38
Hedge, wire fence, railing
245
95
38.8
14
Water tower, small lake, winch
10
4
40.0
16
Ownership border
235
99
42.1
11
Stream, spring, marsh
20
9
45.0
26
Building under construction
19
9
47.4
39
Bushes, orchard, tree
1758
1015
57.7
21
Country road, footpath
543
375
69.1
13
Fountain and pool
14
12
85.7
25
Private building
249
225
90.4
36
Factory, chimney, factory hut
41
48
117.1
17
Disapproval ownership border
94
120
127.7
46
Patrol station and pump
6
8
133.3
15
Tunnel, bridge, stop
67
108
161.2
40
Tree, forest area, green
house
18
51
283.3
41
Park and garden
-
1
+ 1
22
Under- and top-passage
-
2
+ 2
Table 6. Comparison of feature layers compiled on Quickbird
satellite images in 1. sheet
In field control application, the control of outputs of
compilations belong to aerial photographs has been carried out
firstly because most of the data have been compiled on aerial
photographs. While determining an error during the controls,
brief notes have been taken on the outputs about the errors and
then these errors have been controlled on the outputs of
compilations belong to high resolution satellite images. The
attributes of features controlled and the errors determined have
been investigated on laptop and lastly taking all these data into
consideration, it has been tried to evaluate the compilations.
Because of long time interval between images and field controls,
some difficulties have been encountered in finding and
detection of features on Gôlbaçi region which is growing very
quickly.
4.4 Feature Compilation Assessment Results
Using high resolution satellite data, the feature types that are
required for 1:10.000 to 1:50.000 scale mapping could be
satisfactorily identified and captured. In some cases, features
required for larger scale mapping (e.g. roads and woodland
boundaries) could also be identified. But as may be expected, it
is impossible to distinguish the narrow linear features (such as
electricity transmission lines, shapes of buildings, boundaries,
walls, fences and hedges) on satellite imagery. A combination
of panchromatic and multispectral imagery can help to
differentiate between vegetation and artificial features (e.g.
between hedges and walls) but in general the imagery is
unsuitable for the capture of these narrow linear features
(Holland and Marshall, 2004; Holland et al., 2006).
Feature compilation assessment results show that high
resolution satellite images couldn’t reach to the level of aerial
photographs in determining/identifying of small features yet. As
a result, concerning compilation applications, we can say that;
■ The number of features compiled from Quickbird and
IKONOS ortho-images was approximately equal and we
determined that the nearest values to the aerial photographs was
obtained firstly in polygon layer (% 63 - 65), secondly in point
layer (% 57 - 64) and lastly in line layer (% 46 - 50).
■ Quickbird orthophotos showed better performance in line
layer and IKONOS orthophotos have shown better performance
in point layer.
■ The features which were almost not compiled at all in
high resolution satellite images (% 0-10) and acquired in aerial
photographs are; telephone and electric poles, borders, rocks,
stony and sandy places, lean-to roofs and pavements. The
features compiled in minimum number (% 10 - % 40) are;
slopes, natural splits, telephone and electric poles, water wells,
canals, transformers, trees and forest area. The features
compiled in number of % 40 - % 70 are; streams, springs,
hedges, railings and walls, tunnels, bridges, fountains and
bushes. And the features compiled in best number (% 70 -
% 100) compared aerial photographs are; country roads,
footpaths and single buildings.
As an overall assessment for field control applications, we can
say that the operators have had some difficulties in determining
and identifying of some features existing in high resolution
satellite images. These features are; water wells and
transformers taking place in every private country house,
communication and electricity transmission lines in dense
residence areas, electric/illumination poles, wire hedges, small
huts and lean-to roofs. And these results indicate that high
resolution satellite imagery can be used to identify topographic
changes for both large- and small-scale mapping, even if this
imagery cannot be used as a source of direct topographic data
capture (Holland et al., 2006).
5. CONCLUSION
In summary, it can be said that;
■ IKONOS-DEM can be used instead of
photogrammetric DEM produced from 1:16.000 scaled aerial
images and the GCP quality which depends on well spread
distribution and easy recognition is as important as the number
of GCP’s.
■ When using direct sensor orientation parameters,
IKONOS images have better accuracy than Quickbird images.
In addition, systematic errors have been observed in the
easting/north easting (across track) direction.
1233