International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
  
quality analyses of the generated building layer the digitized 
structures have been compared visually with digital 1:1000 
maps. There are some positional differences between the 
buildings which stem from digital 1:1000 maps and digitized 
KVR-1000 image. The discrepancies between both layers do 
not show systematic properties. These discrepancies are rather 
randomly distributed over the entire image. Some locations 
have been choosen for a close inspection.One can take the 
Figure 3 as an example. The left hand-side of the figure shows 
discrepancies between two layers at a location where two layers 
fit each other quite well. The differences lies in the range of 0.5 
and 5 m. But the deviations between two layers are not 
homogeneous in the entire image. The right hand-side of the 
same figure shows two other buildings located 1 km distance 
from the buildings at the left hand-side. Here the disagreement 
between both layers vary about 6 to 13 m. Although an 
accuracy of +4.5 m has been achieved after the transformation, 
this result is not representative in the entire study field. For the 
transformation, generally well-agreed points were selected in 
the cropped image. For more reliable interpretation, 
transformation, between pixel coordinates of selected points 
and their GPS surveyed coordinates has been made. Affine 
transformation has resulted in nearly £12 m accuracy. The 
points used for transformation cover the whole image from 
which our study field has been cropped. Hence, transformation 
for the whole image is more representative than the 
transformation carried out in our study field. 
he 
Discrepancies: 6 — 13 m 
      
Discrepancies: 0.5 - 5 m 
Figure 3. Discrepancies between both layers 
As mentioned before, KVR-1000 orthoimage was taken in 2000 
and the digital 1:1000 maps were generated in 1997. So, there 
are temporal differences between two materials. Although it is 
not possible to reach the geometric accuracy of 1:1000 maps, it 
is clear that determination of new buildings and roads is 
possible by using KVR-1000 orthoimage. For example, two 
centre lines of two different roads have been digitized. One of 
them can be seen in left hand-side of Figure 4. This line is 
located in both the 1:1000 map and the image. The second one 
is not located in the map but is detectable in the image easily 
(see right hand-side of Figure 4). 
  
Figure 4. Digitized center lines of roads 
The operator should accommodate the objects in the image 
during manual digitizing process using radiometric changes in 
the interest area. The contrast of image and effective pixel size 
help this achievement. So, the grey value profile analysis 
should be carried out in order to determine the effective pixel 
size of the image. 
   
   
a 
Average Differences Average Differences 
1.5.96 * L +; 1 68 L 2" 
2 100” 2 6 2.66 2 ul. 
3-105 * 3 3 67 3 
4 105° 4 -1* 4 69 4 4* 
5 104 8. 35 8.73 84} 
6 109 6. 13. * GO 77 6 8 
2-421: * 7 2 x. 7 85 713 
8 143 8 21 "98 190 8 24 
9 164 9:12:95 9 |24 9 24 
10 176 19 > 10 148 10 15 
[1 181 s ]i- 5 11 163 1117 
12 186 12 1* 12 180 12 9 
13 187 13-17 13 189 13 6 
14 186 * 14 -1* 14 195 14 1* 
15 185 15 196 
C d 
Average Differences Average Differences 
] 132. 1 -1* 1,1307 ] -4* 
2 131 2-2" 2 126 2 2: 
3 129 3-5" 3 124 3 0* 
4 124 4 -3* 4 124° 42 
star” STE 5 126 SL 
6 128 6 22 6 125 6.37 
7 150 7 36 : 7 128 7 14 
8 186 8 25 e 8 142 8 29 
9 211 9/77 957171 9 36 
10218 ; 10 0 10 207 10 22 
11218 * ]1.2.* 11229 11.5 
12 220 12 6 12 234 12 -1 
13 226 13 4 13253 13 0 
14 230 3 14 2 2 14 233 14 1 
15232 à 15 234 
Figure 5. Grey value profiles of different edges 
Four different constructions at some selected locations can be 
seen in Figure 5. At the constructions the edges depicted in 
green are selected. The aim of the following analysis is to 
determine contrast of edges, especially between white and dark 
areas. The dark areas can be the grass, water or ground, and the 
white areas can be road or roof of the buildings etc. The edge 
analysis consists of taking profiles along the edge itself. The 
profiles are chosen perpendicular to the edge. For all profiles, 
mean values of the corresponding profile points are determined. 
Thus a graphic showing the trend of average values is obtained. 
The differences of grey values between adjacent points are built 
and can be seen visually. Both graphics can be seen underneath 
the cropped samples in Figure 5. After edge analysis, it is 
expected that the contrast should be very sharp like in object 
  
Inté 
Sp 
nc 
lie 
th: 
th 
an