ints are pro- 
lane, and an 
ed using the 
grid cell. A 
lap to repre- 
er extracting 
range points 
ty poles, ir- 
r differences 
| of trees al- 
image of the 
e points that 
e features in 
ed to model 
nent of trees 
1angle is de- 
al is pointing 
es from each 
n range (see 
leasurement, 
ositions and 
onding strip 
n in the strip 
ie change of 
ra to the ob- 
nerate a tex- 
> images are 
erated using 
se and has a 
is conducted 
ie images of 
'ately, where 
pixel is cal- 
that is made 
ted. 
generated in 
ling the line 
] plane at al- 
und surface, 
ind #5 have 
| by those of 
each vertex 
ge beams to 
ion between 
hetic colors, 
e range points 
ric model of 
Course Length : 1.57 2km A h 
Vehicle Speed : 20- Ak Ii 
va 7 pA AL 
© 2 qu "E 
X uu b= 2 À am 
(Ww or 
bus 
ge 
Gm 
7 qu WM sr 
       
   
F- 
Figure 5: A map of the testing site 
6 EXPERIMENTAL RESULTS AND DISCUSSION 
An experiment is conducted in a real urban out-door envi- 
ronment, KIRIGAOKA apartment complex in Tokyo. The 
measurement vehicle run a course about 1600m at a speed 
of 20km/h~40km/h. A map of the testing site and the ve- 
hicle trajectory is shown in Fig.10. Over 7000 range scan 
lines and 30000 line images are measured by each LD-A 
and line camera respectively as the vehicle moves ahead. 
All range scan lines and line images are geo-referenced 
using the calibration parameters and navigation data. Tex- 
tured CAD model of KIRIGAOKA apartment complex is 
reconstructed by first extracting geometric urban features 
using the integrated model of range points. In this experi- 
ment, there are totally 7,613,502 range points from LD-A 
measurement (see Fig.8(a)). 155 pieces of vertical building 
surfaces are extracted as shown in Fig.8(b). Texture map- 
ping is conducted subsequently using the geo-referenced 
line images. Two perspective views of the textured CAD 
model are shown in Fig.8(c,d), with their viewpoint from 
the ground and on the air respectively. All the processing 
are conducted in a Windows PC with a Pentium III Proces- 
sor, CPU 500MHz, RAM 256MB. Time cost of the total 
processing is about 3~4 hours. 
7 CONCLUSION 
In this paper, a novel method is presented to efficiently 
generate textured CAD model of urban out-door environ- 
ment using vehicle-borne laser range scanner and line cam- 
eras. Both laser range scanners and line cameras are syn- 
chronized with a GPS/INS/Odometer based navigation sys- 
tem, so that the range and line images that measured by dif- 
ferent devices can be geo-referenced to a world coordinate 
system. Generating a textured CAD model of urban out- 
door environment is conducted in two procedures, where 
range data are first exploited to generate a geometric model 
of urban features, line images are then projected and re- 
sampled on the geometric model. Through an experiment, 
itis demonstrated that urban out-door environment can be 
reconstructed with high automation and efficiency. 
ACKNOWLEDGEMENT 
The authors would like to express their appreciation to Asia 
Air Survey Co. Ltd. for their long-term cooperations. 
REFERENCES 
Bosse, M., D.De Couto, S.Teller, Eyes of Argus: Georefer- 
enced Imagery in Urban Environments, GPS World, pp.20- 
30, April 2000. 
Chen, Y., Medioni,G.: Object Modelling by Registration of 
Multiple Range Images, Image and Vision Computing, Vol. 
10, No. 3, Apr. pp.145-155, 1992. 
Collins, R., A.Hanson and E.Riseman, "Site Model Ac- 
quisition under the UMass RADIUS Project," Arpa Image 
Understanding Workshop, pp.351-358, November 1994. 
Debevec, PE, TaylorCJ., MalikJ.: Modeling and 
Rendering Architecture from Photographs: A hybrid 
geometry- and image-based approach, SIGGRAPH, pp.11- 
20, 1996. 
El-Hakim, S.F., Brenner,C., Roth,G.: A Multi-sensor Ap- 
proach to Creating Accurate Virtual Environments, JSPRS 
Journal of Photogrammetry & Remote Sensing, vol.53, 
pp-379-391, 1998. 
Gruen, A.: TOBAGO - a semi-automated approach for the 
generation of 3-D building models, ISPRS Journal of Pho- 
togrammetry and Remote Sensing, vol.53, issue 2, pp.108- 
118, 1998. 
Higuchi, K., Hebert,M., Ikeuchi,K.: Building 3-D Models 
from Unregistered Range Images, Graphical Models and 
Image Processing, vol.57, No.4, pp.315-333, 1995. 
Konno, T, et al: A New Approach to Mobile Mapping 
for Automated Reconstruction of Urban 3D Model,proc. of 
Int. Workshop on Urban Multi-Media/3D Mapping, 2000. 
(CD-ROM proceedings) 
Sequeira, V., et al.: Automated Reconstruction of 3D Mod- 
els from Real Environments, ISPRS Journal of Photogram- 
metry & Remote Sensing, vol.54, pp.1-22, 1999 
Shum, H., Ikeuchi,K. Reddy,R.: Virtual Reality Model- 
ing from a Sequence of Range Images, Proc. IEEE/RSJ 
Int. Conf on Intelligent Robots and Systems, pp.703-710, 
1994. 
Stamos, L, Allen,P.K.: 3-D Model Construction Using 
Range and Image Data, CVPR, pp.531-536, 2000. 
Uehara, M., Zen,H.: From Digital Map to 3D Map: Cre- 
ating 3D Map by Motion Stereo Utilizing 2D Map, IAPR 
Workshop on Machine Vision Applications, pp. 592-595, 
2000. 
Zhao, H., Shibasaki,R.: Reconstruction of Textured Ur- 
ban 3D Model by Ground-Based Laser Range and CCD 
Images, IEICE Trans. Inf.& Syst., vol.E83-D, No.7, July, 
pp. 1429-1440, 2000. 
Zhao, H., Shibasaki,R.: High accurate positioning and 
mapping in urban area using laser range scanner, proceed- 
ings of IEEE Intelligent Vehicles Symposium, pp.125-132, 
2001. 
—123—