International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
3. EXPERIMENT
3.1 Simulator
We implemented the simulation program with C#. For the fast
access to the input data which are object models as formatted in
B-rep, we adopted the 3D spatial database called “PostgreSQL”.
It can store the data and offer more effective 3D spatial
indexing and querying processes. In the simulation operation,
OpenGL with thread technique allows the visualization module
to display the process to generate the simulated point cloud in
real time. Figure 17 shows the user interface of our lidar
simulation software. It mainly consists of three areas for
parameter setting, log message and data visualization. The
parameters setting enable a user to input the system parameters.
A user can specify almost all of the parameters related with a
lidar system, vehicle movement, geometric errors and signal
noises. The visualization area, right of Fig. 1, shows the
wireframe of the 3D city models stored in 3D DBMS.
3.2.2 System Parameters: The scenario of data acquisition
is as following. The airborne laser scanner moves from (-15, 0,
500)m to (15, 0, 500)m with the velocity of 100m/s. The
scanning pattern is determined as zig-zag with scan angle of
Sdeg. Pulse repetition and scan rate of laser scanner are 5kHz
and 0.1kHz respectively. Though there may exist geometric
errors of lidar system, we didn't consider these errors in this
experiment.
The divergence angle of laser beam is determined to be 5mrad;
its size of footprint is of 2m@500m. And we set the pulse width
of Ins and sampling frequency of 10Ghz. Last, we employed 25
(5 by 5) sub-beams to simulate return pulses.
3.3 Simulation Result
After geometric simulation, the 1,550 intersecting points, which
encounter with facets of target, is generated (Figure 19). And
Figure 20 shows the 1,797 simulated points data. The number
of the simulated points is due to detection of multiple returns by
CFD. Waveforms of several locations are shown in Figure 21. It
is confirmed that all of them have return pulses of both target
and ground. But according the shape of intersecting facet,
echoes are slightly different from each other.
Pid iia iiiiigd
HHH
IRRERBHARBABREAR
"EPBHHHBHHS (0PHHHH
EIU c
& OUI IC D UD
7 DHHBHHHun THEE
SUBRHBHIHBERS gf (HHH
SESSRSUERSEREHEESEHZ
SS298P222P0:28529221/
SHRHEHHHHHS
Figure 17. Implemented lidar simulation software 5$ 11 41111711 HIE:
-15 -10 -5 0 5 10 15
3.2 Input Dat xim
ou Jam Figure 19. Intersecting point cloud computed by geometric
3.2.1 Target Model: The proposed method has been tested simulation
with a complicate 3D target model of a kind of fighter. It
consists of 1,257 facets; the length is of about 18m; width is of
about 12m. 1 Üii 1111 :
THEE
\ iii paras : :
e E ui B Hi dodi | i
E > Hu CRUE
: ibi
01-2$-$-$-1 $1'$ $12 1-121422 X
A = HHHHBHIE
FS PIX ; iut 15 di aE
A a xml
Figure 18. Target model as a input data of simulator
521
Figure 20. Final product of simulation
