The stereo-orthophoto contours were compared with contours plotted 
from the original photographs on an A7 plotter. Profiles approximately 
normal to the contours, plotted from an average of 13 and 9 contours 
for the 1:2500 and 1:5000 mapping scales respectively, were used for 
this purpose. Root mean square differences between the A7 and the 
Stereocompiler profiles are given in Table II. 
  
  
  
Table II 
Contouring errors in stereo-orthophotos 
Mapping before correction for after correction for number |aver- 
Systematic displacement systematic displacement 
scale n. ati| m at m m at m at m of age 
z zZ zZ z z CB 
terrain jorthophoto oy terrain [orthophoto 0, points |slope 
oo H oo H 
scale scale scale scale 
1:2500 1.0 m 0.4 mm 10.66 07 0.5 m 0.2-mm-- |0.533 e, 90 6° 
00 00 > 
1:5000 1.5.m.]|.0.3.mm. {0.49 0 1.4:m 4.0.5. mp 10,49. 9/ 72 5 
Jaa OO 
  
  
  
  
  
  
  
  
  
  
  
It was found that particularly the 1:2500 orthophoto contours were 
subject to a systematic displacement. Because of the limited plotting 
area of the prototype Stereocompiler used in this experiment, the 
stereo-orthophotos had to be divided into sections in order to be ac- 
commodated in the Stereocompiler. Small orientation errors of these 
individual sections resulted in some systematic deformation and the 
displacement mentioned above. Since this source of error was due to 
the experimental type of equipment used, the individual profiles were 
corrected by vertical shifting of the data. The figures thus obtained 
are a better indication of the performance of the stereo-orthophoto 
system once more suitable equipment is available. The relative accuracy 
of contour lines is not affected by this correction (Table II). 
The Stereocompiler contours were also checked against spot elevation 
readings, made in the A7 stereomodel along the plotted contours. For 
this purpose, the Stereocompiler manuscript was placed on the A7 co- 
ordinatograph and oriented with respect to the stereomodel reconstructed 
in the A7 from the original photographs. Model heights were read along 
the contours at intervals of approximately 1 cm. After correcting each 
individual contour for a systematic shift, and assuming the A7 model 
readings to be error free, standard deviations of m_ = 0.58 ©/ o and 0.46 
0/00 of the flying height were obtained for the 1:2500 and 1:5000 contour 
maps respectively. 
The contours plotted on the Stereocompiler are similar to those plotted 
by conventional photogrammetric techniques’ as far as details of height 
representation are concerned. This is demonstrated in Figures 1 and 2 
which show a corresponding area from the 1:5000 contour map plotted on 
the Stereocompiler and the Wild A7. 
-11-