The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
395
Figure 9 shows the discrepancies between the true bore-sighting parameters recovered using different three approaches are very
parameters and recovered ones. As it can be seen in these two small, and it can be ignored. The simulation data is relatively
figures, if the slope of the planes is very small (e.g., less than 10 ideal compared to the real data. Anyway, the ultimate goal of
degree), the planar patches are almost parallel, and the RMSE the calibration test using simulation data is to confirm the
of the reconstructed coordinates of LiDAR points are very high. feasibility of the methods and detect any possible drawbacks.
It is also seen that for such a case, the derived bore-sighting The next, the author introduces the experiment results from real
parameters are not close to the true parameters. Therefore, it is airborne and terrestrial LiDAR data. The results are relatively
recommended that some of the control patches should have worse than the simulation test, because the quality of this real
slopes that exceed 10 degrees. Moreover, the patches should data is not guaranteed. Especially, the terrestrial data has some
have different orientation in space (i.e., different aspect angles). drift errors in INS navigation data; there was, however, no
available other terrestrial LiDAR data.
Figure 8. Slope of planar patches and accuracies of
reconstructed surface
_ [meter]
parameters [bore-sight spatial offsets]
GiZ
J j J ■ J
(a)
n
2D
SD
10 D ZJD 300 <00 SOD
Sto pe[ degl
90D
—
0
1
I
1
i
I
3
!
1
1
parameters [bore-sight rotational offsets]
mar
nl
n_- n m mi n_B [La
n_@
ID
5D
10D 23 D 330 <0D 90.0
Slooefdeql
Figure 9. Differences between the true and estimated bore
sighting parameters
3. EXPERIMENTS
In the first test, virtual LiDAR data was simulated and used to
confirm the feasibility of the proposed methods. For the test
data, airborne and terrestrial LiDAR systems are considered in
an urban area. The airborne system flew over the area at around
90m flying height, and the terrestrial system mainly scanned
building walls along the road. As shown in
Figure 10, the terrestrial data consists of two strips, which are
left and right sides of the road.
The results from the simulation data are shown in Table 1. As a
result, the re-covered calibration parameters are very close to
the expected values in all of methods, and these results are good
enough to confirm the feasibility of the proposed calibration
methodologies in this paper. The differences between
Figure 10. Simulated LiDAR data: (a) is airborne LiDAR data,
and (b)&(c) are terrestrial LiDAR data.
A X
A Y
A Z
A co
A $
A K
(m)
(m)
(m)
(deg)
(deg)
(deg)
True
Parameters
0.1
0.1
0.1
1.0
1.0
1.0
ICP
0.11
0.98
0.10
1.00
1.00
1.00
Norm[m] :
0.050
0 : 0.063
ICPatch
0.09
0.10
0.09
1.01
1.02
0.98
Norm[m] :
0.054
0 : 0.071
Planar
0.11 0.10
0.10
1.00
1.00
1.00
Patches
Norm[m] :
0.035
0 : 0.051
Table 1. The LiDAR system calibration test using simulation
data (6 bore-sighting parameters)
Table 2 shows the description of used data; the point density of
the terrestrial LiDAR data is about 2.7 times that of the airborne
LiDAR data. The airborne data was captured at a 150m flying
height from the ground-level, while the average range of the
terrestrial data is less than 20m.
Reference Data
Target Data
System Def-ID
Titan Laser 1
Plate# 3
Point density
13.5 points/m 2
5 points/ m 2
System Def-Time
1-Mar, 2007
11-Nov, 2006
Flying height
N/A
150m
Table 2. Description about test LiDAR data
Figure 11 shows the used real data figures; (a) represents the
terrestrial LiDAR data used as the reference surface, and (b)