. Istanbul 2004
| 3. Each flight
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
First | GFirst
+4 1+4
+3 143
) and standard
alt and gravel
ns. Results are
st pulse data.
The use of centres of buildings
The distances between the laser derived building centre point
and the corresponding centre point from the building map were
calculated, as well as the distances between the centre points of
buildings using the first pulse data and the corresponding
buildings using last pulse data. Lines between opposite corner
points of the extracted buildings determined the centre point of
the building. Results are shown in Tables 4, 5 and 6.
Last 6Last
+4 0+4
+3 143
) and standard
it and gravel
ns. Results are
t pulse data.
point area the
-4 cm standard
The decimetre
oint method at
es of buildings
from the five
in section 2.4.
uildings could
re grouped by
^, double slope
onnected ridge
ridge roof type
.g. ridge roof
trip number 3
ed buildings.
Flight strip | 2F- 3F- 4F- SF- 6F-
map map map map map
Min dist. 7 4 9 | 4
Max dist. 161 67 61 S3 69
Mean +Std | 28428 23416 27:11 1812 24x15
Table 4. Distances (cm) between the centre points of the first
pulse mode derived buildings and map buildings.
Flightstrip | 2L- 3L- 4L- SL- 6L-
map map map map map
Min dist. 7 4 6 1 6
Max dist. 73 52 68 63 82
Mean +Std | 25x15 | 23x14 27x14 19x15 28418
Table 5. Distances (cm) between the centre points of the last
pulse mode derived buildings and map buildings.
Flight strip 2F-2L | 3F-3L | 4F-4L | SF-5L | 6F-6L
Min dist. 1 2 | | 1
Max dist. 41 51 68 35 77
Mean=Std 13+10 | 17413 | 12212 | 14:9 18#16
Table 6. Distances (cm) between the centre points of the first
and last pulse mode derived buildings.
Mean distances of the centre points of the extracted buildings
were less than 30 cm for the first and also for the last pulse data
when compared with the buildings on the map. The standard
deviations varied between «11-28 cm for the first pulse and
+14-18 em for the last pulse. Maximum distances were mainly
less than the Toposys Falcon scan line distance, about 80 cm, in
the across track direction. It was difficult to find the exact
boundaries for some buildings because they were partly covered
by trees. This is one reason for the large errors in distances
when using the first pulse data. Distances between the centre
points of the first and last pulse mode derived buildings
coincide better with each other than with the map buildings.
Mean errors (12-18 cm) and standard deviations (49-16 cm)
from one flight strip include errors resulting from the building
extraction modelling and from the different observation modes.
Mean north and east differences between buildings
from different flight strips
0,25
02 | Onorth
Bl east
0,15
0,1
0,05
0
verag&—hrem
?j
oe c «©
-0,05
avera
-0,1 D050020020040020 006003004006005002002004002004006002004006.
00300400600600300300300508500500300400600600300300300500500:
-0,15
Mean differences and standard dev.
(m)
average [hmmm
-0,2 — — ——— —— ed
Compared flight strips
Figure 2. Planimetric errors of last pulse flight strips. 3 and 5
were flown to the southeast and 2, 4 and 6 to the
northwest. 002-003, 004-003, 006-003, 002-005,
004-005, 006-005 are opposite directions.
Mean north and east differences between ridge roof
buildings from different flight strips
0,25
£ 02 | Onorth |
> Beast .
3 015
©
©
© 0,1
£z
S
o
c 0,05 I
£z
©
o 0 J i
g 1
= 0 19 2 2 9/22 2 2 2 2
© -005 15 FW FE 8 0.5.6 ow ow. o0 ®
o © © © © QO o
€ $8.6 6 6G &
© -0,1
£z i
S D050020020040020040060020080060050020020040020604006002004006
= -0,15 p03004006006003003003005005005003004006006003003003005005005
Compared flight strips
Figure 3. Planimetric errors of last pulse flight strips. 3 and 5
were flown to the southeast and 2, 4 and 6 to the
northwest. 002-003, 004-003, 006-003, 002-005,
004-005, 006-005 are opposite directions. Only
ridge roof buildings included.
As mentioned in section 2.1, strips number 3 and 5 were flown
to the southeast direction and 2, 4, 6 were flown to the
northwest direction. We had two cases to compare. Firstly,
flight strips to the same directions were compared and secondly,
also the opposite direction flight strip comparisons were carried
out, Figures 2 and 3. In Figure 3 only the ridge roof buildings
were included because they were the biggest group and easiest
to extract.
In Figures 2 and 3 the first four columns (average values) on the
left are the same flight direction pairs. Columns five to ten
