Eberhard Steinle 
  
  
  
  
  
  
  
  
   
P 
/ ^ f > 
Su lin e EV / NS 
A Te i Eco 
^ We NR 
J Ty EE 
y. 2 / / f ei 2 / 
gy ary / 
Cd Tos di 
d i ( / 
  
  
  
Figure 7: Recognition of 3D objects as building hypotheses; left side: original first pulse DEM, 
original contour line of separated building including disturbing object, same Douglas-Peucker filtered; 
right side: first pulse DEM after vegetation reduction, resulting contour line (Douglas-Peucker filtered) 
:3.2 Examples 
Several different data sets had been investigated concerning building reconstruction by means of the above mentioned 
approach. In a first attempt, the data acquired by first pulse mode were preprocessed as described in the previous 
section, while last pulse data remain in original status. Additionally, for experimental purposes the first pulse DEM was 
introduced without any preprocessing, i.e. without elimination of disturbing objects like vegetation. As examples two 
buildings in different environments are chosen, both located in our test area Karlsruhe. The presented DEMs were 
acquired by the TopoSys laser scanning system. The flight for data acquisition in first pulse mode was carried out in 
summer of 1997, the one for last pulse mode in the beginning of 1998. Both data sets were georeferenced and 
resampled to a grid of Im x Im. In the DEM roof planes were segmented and contour lines and edge points were 
computed by intersection process which leads to wireframe models. In Figure 8 the results for the first test building, 
using first and last pulse mode, are shown. 
  
  
  
  
  
  
  
  
  
  
  
  
  
Figure 8: Reconstruction process for test building 1; from left to right: laser DEM, segmented roof planes, 
wireframe model in vertical view, same in oblique view; first row contains results derived from first pulse 
data, second row results from last pulse data; in the last row overlay of both results 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 
863