When the superimposed images of the systems 1 
and 3 are corrected for radial scale distortion from 
the model centre the standard deviations are res- 
pectively: 
o, = 6u, 0, = bu and 0, = 
toscale. 
14u, o, = 13u at pho- 
Conclusions: 
1. Table 3 shows the standard deviations for the 
three tested systems. The residual errors 
showed that no systematic errors are present 
when using system 2. 
2. Table 4 shows the standard deviations when 
using for superimposition only the model part 
between the projection centres (Figure 2). 
3. The standard deviations for the systems 1 and 
3 become smaller when the superimposed 
images are corrected for scale distortion. 
3.3. Test B. Accuracy of the superimposed points 
in the field of view 
The problem when determining the superimposition 
accuracy outside the centre of the field of view is, 
that it is not possible to measure superimposed 
points with the help of the measuring mark. The 
reason is that the centre of the field of view is 
determined by the measuring mark itself. 
Therefore the following test procedure has been 
developed. 
One of the gridplates in the analytical plotter is 
replaced by a dense line grid, which has been 
drawn with an automatic plotting table and has 
been reduced photographically. 
The distance between the gridlines is 200 microns 
and the width of the lines.is 10 a 20 microns. 
For recognizability the centre line of the grid has 
been plotted thicker. The standard deviation of the 
grid points for an area of 2x2 cm is 2 à 3 microns. 
During the test the centre line of the grid is super- 
imposed and by changing the position of the meas- 
uring mark the superimposed line together with the 
gridline is moved in steps of 2 mm to the edge of 
the field of view. The movement is executed both 
in horizontal and vertical direction. 
The deviations are observed with the help of the 
dense linegrid by counting the number of grid inter- 
vals and interpolation between the grid lines in units 
of 20 micron (See Figure 3). 
  
Figure 3. Determination of superimposition devi- 
ations. M = measuring mark; A = superimposed 
line; B = grid lines; d = deviation. 
It is assumed that observed deviations are caused 
by the lens system and the screen and so will be 
present in the same way through the whole stereo- 
model. Therefore the measurements are executed 
only in the centre of the stereomodel. To determine 
the influence of zooming on the accuracy the test 
has been executed with different zoom settings. 
If the used superimposition systems are not cali- 
brated in an optimal way the following systematic 
effects can appear: 
1. A constant systematic deviation in x and y 
direction.For the systems of Zeiss and Leica- 
Kern it is possible to correct for this effect by 
moving the superimposed image. For the sys- 
tem of Intergraph this is possible by moving 
the screen or by changing the direction of the 
cathode ray. 
2. A rotation of the superimposed image with 
respect to the photo. This can be caused by a 
rotated position of the screen. For the three 
used systems adjustable screws are present to 
rotate the screen. 
3. Different scale factors in x- and y-direction. 
This effect can be eleminated by tilting the 
screen. 
4. A constant scale-diffference. By moving the 
screen further away or nearer by this effect 
can be eliminated. A limiting factor in this case 
is the range of the depth of field. 
  
NN [7 
tt A 
—" { dai 
ed NN 
das | 
  
  
  
  
ut uta 
eu ue 7A 
wA eur Lun uo ur 
1 
/ 
z 
x 
  
  
  
  
Figure 4. Error patterns of the systems 1, 2 and 3 without correction for systematic deviations. 
Vector scale: 1 mm = 40 micron. 
566