-35 - 
* e The errors obtained in both tests closely agree in magnitude; 
(for the New Guinea test Ay is slightly larger, as may be 
expected due to the huge elevation differences present). 
Interesting is also, that either a polynomial fit or a graphi- 
cal interpolation to some control establishes 1 : 250,000 
mapping standards in position. 
The New Guinea test results in maximal errors in the negative 
of 2: 3 mm in yı and dp Temm in x" scale V5: 300,000). These 
can be reduced by fitting to control. Standard deviations for 
x' and y' are then in the order of * O 2 mm in the negative. 
This closely corresponds to values published by Bosman, Clerici, 
Eckhart and Kubik of + 0,19 mm in æ and + 0,18 mm or 0.24 mm 
in y for a scale of |] : 250,000. 
® € If radar equipment should meet the design goal of 1 =: 50,000 
mapping as stated in [30] accuracies and resolutions of less 
than og = £510 m'are required. They could only be obtained by 
recording of orientation data. 
  
8. Restitution of Sensor Imagery 
The geometrical model presented in section 3 provides the pos- 
sibility for an analytical restitution of sensor imagery. 
  
As was done in the tests (sections 6 and 7) the image coordi- 
nates of discrete points had to be measured in a comparator 
or stereocomparator, Analytical restitution possibly provides 
the most accurate type of restitution, however it fails to 
produce continuous lines, essential in mapping. 
For this purpose analog restitution by plotters has been 
  
applied customarily in photogrammetry. Of the present photo- 
  
grammetric instruments only few are capable of evaluating 
sensor imagery, those that may be termed analytical plotters. 
- 96 -