International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004 
  
The relative accuracy between the manual and the semi- 
automatic mapping is presented in the bottom row of the table. 
In order to compare the two mappings, the accuracy of the 
manual mapping should be calculated. This absolute accuracy 
(including orientation errors) can be evaluated according to 
Lobanov (1984) using Eqs. 8, 9, where: m is the image scale 
(1:41,892), mq is the accuracy of the photogrammetric 
measurements (10 4m), /'is the focal length (153.941mm), and ^ 
is the base line (92mm). However, in the current research, both 
mappings were performed by using the same model solution, 
thus it is the measurement accuracy, rather than absolute 
accuracy, that was calculated using Eqs. 10, 11. The semi- 
automatic mapping accuracy can be calculated according to the 
manual accuracy and the relative accuracy using Eq. 12 and the 
results were Mx = + 0.57m, My = + 0.46m, Mz = + 0.98m. 
E - Eg PE. 
d i LR 
SA e NA 
   
P 
A D 
P: 
  
   
T 
EN onus 
   
    
Lr al 
AN Lua 
XP ass 
Eyes 
M oed 
   
MAREM, 
T PE 
  
x 
X 9 
= ES E 
    
arre 
LUN 
S 
    
    
   
EB mi & Lou 
a 
29 a ou - a 
  
  
  
CM 
  
  
   
  
     
   
     
      
        
    
      
  
  
Br. y De ph 
v SAR, CER & 
d 
   
S. CONCLUSIONS 
As can be seen, the planimetric accuracy of the semi-automatic 
extracted buildings was Mx == 0.57m and My =+ 0.46m. The 
altimetric accuracy was Mz = + 0.98m. 
The results show that using the developed semi-automatic 
approach has achieved numerous advantages, namely: 1) 
extracting 3D building contours from medium scale aerial 
images for mapping or updating the building GIS layer; 2) no 
limitations on specific building shapes or models; 3) this 
approach enables mapping a high percentage of the buildings; 
4) mapping procedure is carried out within a nonstereoscopic 
environment and without using 3D spectacles; 5) using this 
approach increases efficiency, is more economical and reduces 
the building mapping work; 6) the operator can identify at a 
glance which buildings can be mapped by this approach. 
@ = 378 WE SE, Ry) VW MS à Vr ig RN x. 
   
  
  
  
   
  
   
   
    
   
  
   
  
   
     
  
  
  
  
   
  
Figure 2: North Tel Aviv - left and right image building extraction. 
  
     
   
    
   
NA 
her dH Sh 
RT iibris 
SY mane 
  
Figure 3: Central Tel Aviv - left and right image building extraction. 
778 
  
Se 
Fi 
In 
Ir 
th