Gruen, Armin 
  
has been developed for close-range applications, but Streilein (Streilein, 1999) has shown that it may also be used for 
house extraction from aerial images. Figure 3 shows the structure of the algorithms and the data flow. The CAD model 
is used a priori to guide the measurement process and a posteriori to model and store the results in structured form. A 
detailed description of the CAD-based 3-D feature extraction procedure is given in Streilein, 1994, 1999. Figure 4 
indicates several stages of the iteration procedure with an example from close-range photogrammetry (Otto-Wagner- 
Pavillion, Vienna). 
a» 
^ 
  
      
     
  
uliy cem single frame f Semi- 
automatic LS feature extraction « automatic j 
{points, corners, lines) 
  
  
  
  
  
  
matching, monoplotting 
Y 
  
  
  
  
  
v Mame ul qe SU daB simultaneous M E 
: A P vector fields + PONVOE T ae feature extraction 
| e pointclouds ' process with integrated camera model 
  
  
  
  
  
topology builder 
, CAD object à 
house, roof, road | | 
. computergraphics 
   
   
       
  
  
Hd multi-trame. 
feature extraction 
manipulation | | - (points, lines, open polygons, 
visualization | | iteratively closed polygons, planes) 
  
  
  
animation refined 
Y 
3 | Object Builder $i 
y 
i Geometric, Topolouic & Semantic Model I 
  
  
  
Ÿ 
Approximate 3-D Positioning a p 
Y 
2-D Feature Extraction Cl | 
Multiple Kay ts Intersection ] © | 
Y 
Bundle Adjustment 
| Geometric, uu & Semantic Model | 
Y 
Conversion via LISP 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
   
43 ... Human operator interaction 
QD... QD Processing loops 
Figure 3. Flowchart of semi-automated CAD-based 3-D object extraction 
  
312 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.