location and thus are distributed in space at random a noisy 
  
  
  
    
  
  
  
  
    
  
  
  
   
  
  
  
  
  
  
  
  
  
  
    
  
  
   
   
   
   
    
image is produced. 
5. Comments on the Image Outputs 
Figures 2, 3, #, and 5, 6, 7 refer to the areas ^ and 2. The 
letters a - f correspond to the channel combinations 4e 5, 
Ho, 4a 7, 506, 597, eT. 
6. Conelusion, Outlook 
With some selected examples it was intended to demonstrate 
how desired information on linear dependencies can be ob- 
tained by means of an appropriate combination of simple time- 
saving procedures. E.g. the computing time on the large ca- 
pacity computer TR 440 of the "Gesellschaft für Mathematik 
und Datenverarbeitung (GMD)" (Society for Mathematics and 
Data Processing) at Darmstadt amounted to 2 minutes each for 
the areas 1 and 2 with 86 400 pixels each, while for the 
area 3 with 1.6 millions pixels 20 minutes were spent. It is 
planned to apply this procedure also to test the colour sep- 
aration qualities of the scanning system P 1700 (of the Pho- 
togrammetry Division of the "Institut für Angewandte Geodä- 
sie (IfAG)" (Institute for Applied Geodesy). 
7. Acknowledgement 
Finally, I want to thank Mr. Knaus (Wissenschaftlicher Ober- 
rat in the IfAG) for the establishment of the computer pro- 
gram used for this examination. 
8. Bibliography 
[1] Niemann, H.: Methoden der Mustererkennung, P.147 ff. 
Frankfurt 1974 
ro] Wolf, H. : Ausgleichungsrechnung nach der Methode 
der kleinsten Quadrate, p.!19 ff. 
Bonn 1968