The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Pol. XXXVII. Part B5. Beijing 2008 
558 
results are mean values for the five reference spheres in the four 
scans of different loadings. For the sphere targets made of 
Styrofoam with a diameter of 12 cm, the mean absolute error 
and the standard deviation correspond to the results for the 
reference spheres. 
For the registration of the 3D-point clouds, all the 3D-point 
clouds were registered into the coordinate system of the initial 
scan. The registrations were performed with the software 
Cyclone by Leica Geosystems AG (http://www.leica- 
geosystems.com). The registration quality is specified by the 
mean absolute error and the RMS (Root Mean Square). The 
mean absolute errors for the three registrations were 0.6 mm 
(initial-Pl), 0.6 mm (initial-P2), and 0.5 mm (initial-P3). 
Furthermore, the RMS was calculated to 0.6 mm for each of the 
three registrations. 
3.2 Deformation analysis between different epochs 
As mentioned above, the deformation analysis of the TLS data 
was performed both area-wide and discrete. The latter was 
performed by using the sphere targets on the object. 
The area-wide deformation analysis required a filtering of the 
3D-point clouds by eliminating outliers and undesired points. 
Furthermore, the scanning section to be analysed was restricted 
to a 10 m by 20 m area (Figure 6). The filtering of the 3D-point 
clouds was performed with the software Geomagic Studio by 
Geomagic Inc. (http://www.geomagic.com). Hence, the area of 
the road surface was filtered identically for each scan. Filter 
algorithms were run for an automatic detection of outliers, a 
reduction of the point spacing (40 mm) and a smoothing of the 
3D-point cloud by a free form filter. The average distance, 
which the 3D-points were moved, was calculated to 1.6 mm 
(mean value of all four scans) and the standard deviation of the 
residuals to 1.3 mm (mean value of all four scans). 
The deformation analysis was performed with the software 
Geomagic Qualify. The 3D-point clouds of the different loading 
situations were compared to the initial situation. The residuals 
were calculated as the shortest distances from the scan points to 
the initial surface which was modelled by triangulating the 30- 
point cloud. 
In Figure 6, the residuals of the scan points of the loading 
situation P2 to the initial situation are shown. Deflections of the 
outer side of the cantilever slabs are detectable. The maximum 
deflection is around 20 mm. The results for the comparison 
between initial situation and loading situation PI are similar but 
the sizes of the residuals are smaller. The influence of the 
different loadings on the resulting deflections can be clearly 
distinguished. 
Table 1 summarises the results of the comparison between the 
different loading situations and the initial situation. Differences 
between the results of the different loading situations are clearly 
detectable. However, the maximum positive and negative 
residuals have to be looked at with care due to the fact that 
these values can be influenced by outliers which could not be 
detected during the filtering process of the 3D-point clouds. 
Besides the area-wide deformation analysis, a discrete 
deformation analysis was performed by the 13 sphere targets 
which were arranged around the loads on the cantilever slabs 
(cf. Section 2.5). The target sphere centre points for the 
different loading situations were compared with the initial 
situation. Figure 7 shows the vertical displacements (Az). The 
largest residuals can be detected for the targets 104 and 204 
which were located close to the loads on the outer side of the 
cantilever slabs. The mean values of the residuals are calculated 
to -0.2 mm for the differences between situation PI and initial 
situation (standard deviation: 1.6 mm), -1.7 mm for P2 and 
initial (standard deviation: 2.7 mm), and 0.4 mm for P3 and 
initial (standard deviation: 0.5 mm). 
o 6 W 05 o 
o b b o b 
deflection 
Figure 6. Deflections of cantilever slabs detected by 
terrestrial laser scanner (residuals of the scan points 
of loading P2 to the initial situation). 
Mean 
positive 
residual 
[mm] 
Mean 
negative 
residual 
[mm] 
Max. 
positive 
residual 
[mm] 
Max. 
negative 
residual 
[mm] 
Standard 
deviation 
[mm] 
PI 
0.6 
1.4 
6.0 
10.5 
1.4 
P2 
0.5 
3.6 
4.8 
24.3 
1.9 
P3 
0.5 
0.9 
7.6 
9.6 
0.9 
Table 1. Mean residuals and standard deviations of the 
residuals for the different loading situations as a 
result of an area-wide deformation analysis. 
Figure 7. Vertical displacements of sphere centres between 
different loadings and initial situation measured by 
terrestrial laser scanner Imager 5006. 
In general, it can be said that deformations of the cantilever 
slabs could be detected by the area-wide analysis as well as by 
the discrete analysis. The discrete analysis shows smaller 
residuals which are caused due to the fact that the target spheres