erfect ellipse, 
ellipse is re- 
int to identify 
k in the frame 
lues are taken 
g the ellipse. 
known object 
As this also 
and recording 
e of the signal 
ize is used for 
l. 
ellipse focus) 
ting ellipse of 
is too high 
and measured 
gradients on a 
arting value is 
Xf black target 
from bright to 
such a transi- 
itinued until a 
The center of 
owing starting 
rent. To avoid 
d position, the 
already failed, 
jew positions. 
/ from the ex- 
ried. This pro- 
nts, as for ex- 
scale field is 
corresponding 
) 3 
roughly to the 
the basic grey 
xamined under 
minimum. The 
ined by calcu- 
g values. This 
values, cor- 
ey scale func- 
transition from 
folding has the 
rferences as it 
rge spectre of 
d by the addi- 
ove the detec- 
  
Figure 6: Recognised ellipse 
Identification of Ellipses 
The automatic identification of signals placed on the 
object is an important step towards comfortable usage 
and efficiency during the measurement procedure. This 
kind of correspondence between points and point 
numbers is very significant when the image carries 
retro-reflective target markings. The normal run con- 
sists of image orientation with only a few known 
points and a calculation of the approximate values for 
the remaining object points. Because of the lack of 
natural object information, this manual coordination 
can only be effected by the help of special target 
marks or additional point numbers. 
A new way is the use of target marks with point num- 
bers that can be identified automatically. In this case, 
only the positions of the target marks must be known, 
not their numbers. By means of digital image proc- 
essing, the signal of the point number is recognised 
after having measured the target mark. 
The marks have to fulfil the following demands: 
* Tolerance against interferences 
e Possibility to control the identified point numbers 
e Efficiency of the technique 
*  Possibility of adapting same type target marks of 
different sizes 
Target markings carrying a binary segment code on the 
single segments have proved to be most resistant 
against interferences (segment marks). This kind of 
code uses almost the whole surface of the signal mark 
for point coding and only it's large surface protects 
them from interferences or inaccuracy in the ellipse 
detected before. 
Using a nine-bit-code requires nine segments for the 
bits and one segment marking the beginning of the 
code. With nine bits it is possible to code 512 different 
marks, sufficient for common applications. 
The segments are always divided into an inner and an 
outer part. For segments covering one bit of the code 
both parts are identical. However, the segment repre- 
senting the starting mark differs in its parts. This dif- 
ference allows the identification of the starting mark. 
For this purpose the grey spectre is examined on two 
radiants around the recognised ellipse. The starting 
mark is found when the grey values on the inner and 
the outer radiant differ significantly. Starting at this 
mark, a grey course is recorded on a radiant around 
the ellipse. This grey profile is then examined under 
aspect of bright-dark variation. As the sectors con- 
taining the codes are known by the number of coded 
bits, the centers of these sectors can be emphasised in 
the correlation. This assures a very safe identification. 
    
   
measurement 
ellipse 
inner aM 0 outer radiant 
1 
Figure 7: Segment mark 
The examination of the grey value spectre is effected 
first in the outer part of the segment marks and then in 
the inner part. Only when both evaluations correspond 
the point number is regarded as being recognised. 
The size of the target mark are variable because the 
sizes of the inner circle, the radiant of the entire mark 
and the separating radiant between inner and outer 
segment can be adjusted by the user. Furthermore, the 
number of coded bits can be adjusted in a parameter 
file from the outside. 
REMARKS 
The MIROS measurement system was intended to be a 
semester paper and has meanwhile reached the stan- 
dard of a commercial product. In cooperation with 
Rollei Fototechnic and the international industry, vari- 
ous performance tests proved the flexibility, reliability 
and measurement accuracy of MIROS. 
The successful completion of this measurement pro- 
gram was made possible with the support of Rollei 
Fototechnic. With special thanks to Prof. Dr.-Ing. 
Günter Pomaska and Dipl.-Ing. Günter Suilmann.