ISPRS Commission III, Vol.34, Part 3A „Photogrammetric Computer Vision“, Graz, 2002 
  
  
Figure 3. Extracted 3D roads and road network (the black lines) 
in test site Thun, Switzerland Figure 5. Extracted 3D roads and road network in test site in 
Belgium superimposed on image as black lines 
  
  
Figure 6. Details of road extraction and junction generation in 
Belgium dataset. Extracted roads are shown in black 
lines and outdated roads in white lines 
  
  
  
  
  
  
  
  
  
  
Quality Measures Figure3 | Figure 5 
Completeness 93% 97.6% 
Correctness 96.3% 98.1% 
dx -0.08 0.04 
Mean of dy 0.07 -0.11 
difference dz 0.19 0.27 
dx 0.45 0.57 
RMS of dy 0.31 0.72 
difference | dz 0.62 0.89 
  
  
  
  
  
Table 2. Quality evaluation (in m) of the reconstructed roads 
using manually measured reference data for the 
Swiss and Belgium datasets 
  
Figure 4. Details of road extraction in the test sites in 6. DISCUSSION AND CONCLUSION 
Switzerland. The Extracted roads are shown in black 
lines and outdated roads in white lines. a) Road We presented a knowledge-based image analysis system for 
extraction and junction generation, b) extracted lane road extraction from stereo aerial images. The system has 
border lines and lane centerlines on a highway several advantages over other approaches. It uses existing 
A - 404