In images with a preilluminated reseau (Figure 5), the 
measurement is structured as follows: 
At first, the mesh to be read is focused by the sensor 
and the newly calculated through-light intensity is ad- 
justed for the measurement (compare homogenising 
the digital partial images). When the partial image is 
given into the frame grabber, the sensor is moved to 
the following mesh and the next illumination intensity 
is adjusted. The time, the scanner requires is used for 
initialising the mesh and for measuring the signal. 
  
Figure 5: Picture with preilluminated réseau 
Homogenising the Digital Partial Image 
To assure a fast and safe signal measurement, it is 
necessary to avoid interfering influences by means of 
digital image evaluation or adjustment when snapping 
the digital partial 1mage. These are the prerequisites 
for a fast image identification and for highly accurate 
positioning. 
For accurate measurement, an even illumination of the 
image is essential. An over-exposure would falsify the 
objects relative to the background. If, for example, a 
black ellipse has to be measured, its edges would 
move towards the center in the case of over-exposure 
as the bright background would shine into the ellipse. 
In another extreme case with an under-exposure of the 
image, a black ellipse would not have sufficient con- 
trast against the black background and the edges could 
not be identified. 
The fact that negatives always show an uneven illumi- 
nation and the necessity of a perfect illumination of the 
image for an accurate measurement results in the de- 
mand for an automatic postadjustment of the light used 
for snapping the image. This demand was introduced 
into MIROS routines as follows: 
During the orientation, the through-light intensity, the 
user adjusts for measuring réseau crosses is stored. For 
any later measurement, this value is taken as a starting 
parameter. 
The algorithm for post-adjustment of the illumination 
intensity does not determine the intensity explicitly 
from the grey scale which would lead in the case of er- 
roneous measurement to false light values and to the 
end of the procedure. The grey scales of the signals 
are taken and then, with the example of a white signal, 
the correspondence between the grey spectre and the 
maximum grey spectre value can be determined. 
Should this be the case, the illumination intensity is 
decremented or incremented. With this procedure, the 
illumination intensity always varies around the optimal 
value. If a signal is not measurable, the illumination 
intensity 1s not post-adjusted as no reference values 
are available. 
In contrary to the regulation of different illumination 
intensities as described before, only one illumination 
intensity is regulated for images with réseau illumina- 
tion. In retroimages, the réseau crosses are visible in 
the through-light for orienting the sensor and it is 
therefore possible to effect the orientation of the sen- 
sor and the measurement in the same image. The light 
cannot be focused on the crosses or the ellipse to be 
measured as this would mean that one of the signals 
would always be insufficiently illuminated. For this 
reason, the grey scale values of the background are 
taken for illumination regulation. After the measure- 
ment of the crosses and after the measurement of the 
signal targets, a medium grey scale value is chosen 
from the area in which only background grey values 
are placed. This grey spectre value is regulated on a 
value that makes possible both, the measurement of 
the white réseau crosses and of the black signal tar- 
gets. This procedure assures a sufficient contrast of the 
targets and the réseau crosses to the background. 
Only three important grey values occur in the retro 
images: crosses, signal targets and background. There- 
fore, the capacity of the frame grabber to differentiate 
between signals and surroundings can be used. From 
the original image, two copies are made within the 
frame buffer, one for isolating the crosses, one for 
isolating the signal targets. In retro images, signal tar- 
gets are always black ellipses. This means that the 
significant grey values for measuring these signals are 
always in the lower grey value spectre. For this reason, 
recordings within the screen buffer is always effected 
via a threshold Look-Up-Table. Its purpose is to give 
identical copies of the grey values from 0-50 and to 
take a value of 50 for any others. Thus, bright interfe- 
rences are eliminated and darker interferences are re- 
duced. The only sharp gradients in the image are then 
the edges of the ellipse which can then be found and 
measured easily. A similar operation is effected for the 
réseau crosses: any grey values below the threshold 
value of 150 are set at 150, grey values above this 
value are accepted. To react flexibly on the different 
  
   
  
   
   
    
   
  
   
   
  
  
  
   
    
   
  
  
  
  
  
  
  
   
  
   
   
  
   
  
   
  
  
  
   
    
    
   
   
   
  
   
   
  
   
  
  
    
  
   
   
  
   
    
   
  
        
‘negati 
the ci 
adopt: 
image 
backg 
this r 
réseat 
and in 
grount 
crosse 
crease 
and is 
then e 
light i 
tion. S 
culate: 
mente 
A pre 
illumi 
even 1 
possib 
illumit 
quent 
parts « 
in the 
crease 
can be 
the sc: 
dual c 
illumir 
then t 
happetr 
The d 
avoide 
of spe: 
able to 
to his « 
For st: 
terpret 
genera 
the lv 
structu 
file ar 
shell i 
possib 
the wc 
comm: 
1. Inc 
alle 
prc 
col 
anc