METHOD OF IMAGE ERROR PRESENTATION 
  
A comparison of the effects of various sources of 
image geometry changes by means of root mean square errors 
is customary (e.g. in [44] and [53]). However, these changes 
often cause systematic components not eliminated from the 
r.m.s. errors, In this paper, a different approach will be 
taken. The relative size of various image errors will be 
shown in veetor-graphs. The majority of these graphs was 
attained in evaluating wide angle photography taken at a 
scale 1:10000 over a test area, pre-exposed grids and cameras 
calibration plates. The photography to be discussed was 
taken simultaneously using two reseau cameras equipped with 
a register glass reseau ([72],[69]), a Wild RC8 Reseau camera 
and a Zeiss RMK AR 15/23 camera. 
All sources of changes in measured image coordinates 
which can be controlled using such a reseau are indicated 
in Pig. 1, with heavy lettering. 
For a comparison of the relative size and importance 
of the image geometry changes it is not necessary to specify 
the scales used as long as all graphs are produced using one 
scale for the location of image points and one for the vector 
lengths. 
In most cases, a linear conformal or similarity trans- 
formation of the measured values to the known point positions 
was applied. The reason for this is twofold: firstly, a 
change in focal length to accommodate a scale change of the