It can be seen from Table III that generally a good agreement 
of the results was obtained for each of the two orthophotos produced on 
the same instrument, with the exception of the results from the Zeiss 
Ortho-3. This latter discrepancy was also detected in the Stereocompiler 
measurements. Also the tangential standard error for one of the ortho- 
photos produced on the Ortho-3 is significantly larger than for other 
instruments, which is a reason to believe that the difference in the 
results obtained for these two orthophotos is due to an excessive, irregular 
dimensional change of the original orthophoto film. 
It can be seen from the results in Table III that the tangential 
standard errors are approximately the same for all ‘instruments. The 
radial standard errors are significantly larger and also less uniform for 
different instruments, which indicates that the main error sources 
affecting the geometric quality of orthophotos are indeed shortcomings 
in the differential rectification process such as the size of the rectifi- 
cation slit and profiling errors. 
As expected, the radial errors increase with the size of the 
elementary rectification area (slit) if secondary rectification within 
this area is not used, Fig. 3. 
The beneficial effectof linear transverse slope correction used 
in some conventional slit type instruments such as the Wild OR-1 and the 
Zeiss GZ-1 results in relatively small radial errors compared with the 
slit size, Fig. 3. The improvement achieved probably amounts to 25% - 
302. This is of practical importance since the introduction of slope 
corrections permits the use of larger slit sizes, which in turn considerably 
reduces both the scanning time and production costs of orthophotos. 
The radial standard errors determined for the projection type 
instruments (which do not use slope corrections) such as the Ortho SFOM 
9300, the Zeiss Ortho-3 and, to a certain extent, the Kelsh K320 exceed 
those for the other instruments. This was analyzed further in relatively 
flat terrain for control points located within 0.25 mm (at the scale of 
the original photos) from the scanned profile. Radial orthophoto errors, 
determined for these points, are mainly a result of the profiling errors. 
The averaged values determined for both orthophotos, listed in Table IV, 
corroborate the fact that the relatively large standard errors in Table 
IIT obtained for the Ortho SFOM 9300 and the Zeiss Ortho-3 are indeed due 
to inaccuracies in the profiling operation. It is interesting to note 
that the results for the Kelsh K320 in Table IV are considerably better 
than for the Ortho SFOM 9300, which is a similar projection type instru- 
ment. This could be the result of the elegant solution for profiling on 
the Kelsh instrument which avoids any movement of the heavy projection 
table during the scanning operation. Instead, the operator controls only 
a light reference surface on which both photographs are projected; the 
image is transferred to the stationary film by fiber optics. 
The rectification plane established by automatic image correla-