transformation unit can be quite successful if the deformation 
is regular as for example that shown in Fig. 20b and Fig. 223 
and D. However, local irregularities can be effectively 
corrected only by local transformation involving an image 
point and several surrounding reseau crosses. This was also 
proven in [70] and [71]. 
SUMMARY AND CONCLUSIONS 
  
An attempt has been made to report the developments 
regarding the image geometry and its ehanges since the last 
Congress of the ISP. The main interests during that time 
period have been directed towards problems of the determination 
of lens distortion in aerial cameras and the calibration of 
the entire photogrammetric system up to the measuring stage. 
Such a system calibration provides the best possible 
compromise in the determination of more or less correlated 
orientation parameters. Both random and systematic errors 
arise from a variety of sources. The more important ones 
were discussed and an attempt was made to demonstrate their 
contribution to the overall change of the image geometry. 
The practical value of a system calibration is limited, 
because the ambient conditions will likely be different in 
each calibration. Only for the project within which it has 
been performed will a system calibration be extremely helpful. 
Statistical material on system calibrations may, if it becomes 
large enough, also contribute to the determination of the