International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
4.2 case 2 
Fig. 7 shows the points cloud of another building, which 
consists of 3050 thousands points. Fig. 8 is the projection of the 
building onto horizontal plane. With the condition of DoPP 
>=30, an useful result is obtained and is illustrated in Fig. 9, 
where 23 000 raster grid points are used and a narrow valid 
Fig.7 3D range image of building 
Honzertai Protechon 
root 
  
7 - ae 
ege 
. go. 
e f # fr à 
Y-ans 
  
mar rn tectorum nmn one 
Xu 
Fig9 Projection to horizontal plane as DoPP >=30 
  
  
Fig. 11 3D Model of Building 
  
  
— mmc — —M— 
ü 19 m a 40 £0 oe nN en 9o 
Fans 
  
me i d d 
A { 19 (ü 
region is generated. The stochastic noise into the building are 
all together thrown off; The horizontal outline of the building 
can be obtained by further linear feature extraction and figure 
structure edition, see Fig. 10, and the 3D model of building 
with the height information of building can be rebuilt. Fig. 11 
shows the result with average height of the building. 
  
  
   
  
  
  
o 
Fig. 10 Result of analyses and interactive editing 
5. CONCLUSIONS AND FURTHER RESEARCH 
The vehicle-based laser scanning system is a new technology 
for quick acquisition of 3D information. Ground targets, 
such as urban buildings can be extracted from laser-scanned 
data. One of the most critical issues for the application of this 
laser-scanning technology is the development of efficient 
methods for extracting and reconstructing objects from the 
huge amount of laser-scanned data. 
This paper presented a research development on the extraction 
of urban buildings from vehicle-borne laser-scanning data. An 
operational workflow of the solution was proposed. With 
  
   
aux 
im: 
mj 
ext 
fac 
use 
stri 
det 
Acl 
An 
An 
Bn 
Sur 
Eri 
Sta 
Phe 
Flo 
Tor 
Rac 
pp. 
Ha: 
Air 
Kil 
Kra 
Wo 
Jou 
193 
Li 
Fea 
Infc 
Li ( 
Sca 
Uni 
Lu 
Urb 
Eng