The lenses, adapter rings and converters must have the correct location in relation 
to the optical axis. A decentering can have an influence to the geometric situation. 
A screw off and on again has had a small influence to the location of the principal 
point and also to the distortion. The distortion is not only radial symmetric. Like 
shown in fig. 5 there are in some combinations two centres for symmetry 
overlapping, which can be explained by a decentering of some optical parts. 
!» » 
9 
l 
* 
1 
l 
i 
\ \ 
V' 
\ \ 
J 
T 
s 
N 
N. 
v 
\ 
1 
V 
l 
l 
/ 
/ 
/ 
fig- 
5 
\ r 
r 
p 
- 
V V 
t 
- 
- 
- 
'i* 
* 
*- 
distortion 
of 6006 
* 
yp 
p 
t 
V 
with 
34mm adapter 
/ / 
. 
s 
r 
T 
\ 
\ 
and 
2-fold 
converter 
V 
- 
- 
/ 
/ 
T 
\ 
\ 
\ 
Yi 
V 
• 
\ 
/* 
/ 
r 
T 
; 
\ 
\ 
1142 
VERZEICHNUNG R6006 TELC.+34 > '60.00 u 
5. Video cameras 
The used solid state cameras do have an array of 768 * 512 pixel, each 
approximately 10 pm * 8 pm, in total 6.1 mm * 4.1 mm. This is much less than the 
corresponding photo size. For a comparision of the geometric angular resolution, the 
accuracy relations have to be respected. Based on Rollei-images a standard 
deviation of unit weight oo = +/- 6 pm has been reached. The not optimal results 
mainly have been influenced by the limited depth of focus. With video cameras an 
accuracy of oo = +/- 1 pm is possible. In relation to the accuracy the array size is 
corresponding to a photo size of 36 mm * 24 mm like present at the Rolleiflex 
metric 3003. That means, the information contents of the 6006 will not be reached 
with the used array size. 
The mentioned accuracy of the video cameras has not required any precorrections. 
The geometric situation of the used cameras is excellent. The radial symmetric 
distortion of the different small angle lenses is below 10 pm. 
The aimed accuracy of few microns in object space is limiting the image scale. On 
the other hand the object has a vertical dimension of 0.6 m. Only with a relative 
positioning the requirements can be fulfilled, it is not possible to place a control 
pattern around the object, so beam splitters have been used infront of the video 
cameras (see fig. 6). The location of the cameras will be determined in relation to 
the control pattern and based on this information the object coordinates can be 
calculated corresponding to a location of the control pattern around the object. 
Depending upon the lighting, the control pattern and/or the object can be imaged. To 
avoid a negative influence of the control pattern to the automatic object 
identification, at first the control pattern, after this the object will be illuminated, 
checked again by the control pattern.