B3. Istanbul 2004 
and sidelap (such 
e of a 3-D model 
photos. 
  
xture image of Face 
ter of Imgl have 
selection 
t and pasted from 
TS 
LIDAR data and 
n Tokyo (50-cm 
LIDAR sensor. À 
| application that 
iter. 
cted polygons of 
a sample of 3-D 
ows that texture 
building, as well 
ther sample of 3- 
s, the processing 
'V processor. 
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004 
  
The 3-D model enables visibility analysis from each viewpoint 
of aerial photos. This realizes true orthoimage creation based on 
the re-created DSM from extracted building polygons. Figure 7 
shows a sample of a true orthoimage of Shinjuku area (Oda et. 
al., 2004). 
4. SUMMARY AND FUTURE WORK 
A method for automatic building extraction and 3-D city 
modelling has been presented and demonstrated with LIDAR 
data of Shinjuku and Ginza area in Tokyo. This method can 
se 
  
    
produce 3-D models with texture images and the processing 
time is about 2 hours for Ikm? with a Pentium IV processor. 
This method now cannot express inclined roofs. This will be 
attained by segmentation of DSM which can distinguish 
inclined area as fragments 
.References 
Oda, K., W.Lu, Uchida, O., Doihara, T., Triangle-based 
visibility analysis and true orthoimage generation, ISPRS2004, 
Commission III, PS WG III/8. 
     
  
Figure 5. A sample of 3-D city model processed from LIDAR data and aerial images. (Shinjuku area)