Stephan Heuel 
  
  
  
  
  
  
  
  
  
  
  
  
  
Grouping All Topology Geometrical 
Stage Tests | Selection Approval 
sec. 4.3 1007 | 361 (3602) 98 (027%) - 
sec. 4.4 1303 | 362 (27%) | 219 (60%) iL . 
sec. 4.5 328 (73 (2%) | 53 (72%) p om 
  
Table 3: The table lists statistics for the three grouping stages 
desribed in sec. 4.34.5. The first column lists the number of 
all possible tests at this stage. The second column gives the 
number of selected tests just by using topology, additionally 
the reduction factor is given in percentage. The third column Figure 5: Synthetic example. The left picture shows the 
contains the number of succesful geometrical tests; here, the Set of generated n-corners and the right the set of recon- 
number in brackets gives the percentage of accepted geomet- Structed surfaces from these corners. Note that the top sur- 
rical tests from the accepted topology tests—if it would equal faces of the two polyhedra are detected as distinct though 
100%, the geometric test would be useless. coplanar. 
  
  
Figure 6: Building example 1. The first row shows two out of four images for one tested aerial scene , the 
feature extraction for one image and the Voronoi diagram of the features yielding to neighborhood relations of 
features (cf. sec.2.1). The second row shows the set of generated n-corners (left) and the set of reconstructed 
surfaces (right). 
  
Figure 7: Building example 2. Another reconstruction of an aerial scene with a set of n-corners 
(left) and the found surfaces (right). Note that the long direction in the back of the scene was 
removed because of the bad standard deviation of the angles. 
  
404 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.