International Archives of the Photogrammetry, Remote Sensing
and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
The acquisition results are summarized in Table 2. We scanned
the building at 44 different positions, covering the entire
building facets with a nominal point spacing of about 6 cm and
providing more than six million points.
| 5.96 em |
_38.39m |
6,454,580 |
| 100.5 MB |
Average range
re A Crema eee UM erret
Number of points
_File Size
F
M
-
Table 2. Main specifications of the laser scanner (ILRIS-3D)
5.3 Processing Results
The registration results are presented. The 44 sets acquired at
different positions were integrated using the registration
processes. The integrated set is shown in Figure 7. The
residuals resulted from the registration are partially presented in
Figure 8, where the magnitudes are encoded with the colors.
Most of residuals are within the acceptable range, + 10 mm.
Figure 7. Registered point clouds (Integration of 44 sets)
Figure 8. Residuals of registration
5.4 Evaluation Results
We evaluated the derived polyhedron model resulted from the
acquisition and processing processes by comparing the model
with the original planning model of the building. A floor plan
was extracted from each model, as shown in Figure 9. We
compared the length of every corresponding straight segment.
The histogram of the differences in lengths resulted from this
comparison is shown in Figure 10. The standard deviation of
the differences is about + 37 cm. This magnitude is much larger
than the value (+ 1 cm) we expect based on the specifications of
the laser scanner. This indicates that the differences originate
from other sources than the random errors associated with the
laser scanner measurements. Also, this implies that there are
significant differences between the planning model and derived
model. To clarify which one is correct, we directly measured
three segments (2, 3, 4) using a total station. These results are
described in Table 3. Every difference between that from the
derived model and that from the total station is less than 5 cm.
The values from the derived model are more consistent with the
total station results than those from the planning model. This is
reasonable observations since the planning model is not usually
perfectly kept during the construction of a building.
Furthermore, three segments (14, 16, 34) noticeably show the
difference of larger than 1 m, as indicated in Figure 10. This
must be also due to the construction errors.
TA
Figure 9. Comparison of the floor plan extracted from the
derived model (red solid polygon) with that from the planning
model (blue dashed polygon)
Numer of segments
w
34 16 14
1 [
0 |
-1500 -1000 -500 0 500 1000
Differences [mm]
1500
Figure 10. Histogram of differences in the lengths of segments
988
In
Wi
for
clo
im
prc
sur
ter]
mo
Thi
WO!
Doi
pho
Acl
and
Ren
Bali
and
Ren
Ede
the
Info
Hab
calil
Arc]
Infoi
144-
