camera in used configuration 
focal length principal point 
xp | y» 
6006 + 2-fold converter + 34mm adapter ring 
218 +04 | 0.73 + 04 |1G1+ 08 | 1418 
6006 + 2-fold converter + 68mm adapter ring 
275+03 | 046: 02 |- 62x04 | 171 
6006 + 2-fold converter + 102mm adapter ring 
39220903. ] 075205 |-119517 | 253 
3003 + 2-fold converter + 2 adapter rings 
scale 
  
  
  
  
260 +03 _| 0.54 + 09 [0.33 + 06 | 261 
3003 + 3 adapter rings 
138 £ 0.1 0.92 + 0.5 |1.38 + 0.8 | 1.5:1 
  
Table 2: results of calibration [mm] 
Rolleimetric 6006 and 3003 
After the first test, the camera configuration 
has been dismounted and mounted again to check 
the possibility of restoring the camera system. 
The differences of the second calibration against 
the first have been just at the significant limit. 
Especially the location of the principal point also 
changed the sign. As final result, the fiducial 
center was used as principal point. 
With the fixed parameters of the inner orientation 
the systematic image errors have been determined 
by bundle adjustment with self calibration by 
additional parameters. The systematic image 
errors of both Rolleimetric cameras are quite 
different depending upon the number of the used 
optical elements. The structure and also the size 
is changing. Also after dismounting and mounting 
the systematic image errors are changing. That 
means, the selfcalibration has to be included into 
the standard solution of orientation determination 
and model handling. 
  
  
  
  
  
i Xd : - 2 i 4 
; NN ner > e x i NS 7 = z= AN : = 
| a = Y pére re EX mv : 
fae Nub sd Ne E o Fin NN d 
e pue pet Ea AM. u ud € - - 
b xe ri ipao iens nnm À 
SI Hide ud ocu e ord ba di 
/ NI | | mm . 
| | LE ap ae RTC m bu. AA CN 
1e ++ > | aeu cus ARS ITH Eh a aR 
[N | Ce A i 
| e PURUS i | | 4H mo TA 
i lua ! d > an 
Fig. 4: systematic image errors of the different 
configurations of the Rolleimetric 6006 
Similar results like with the Rolleimetric with the 
stereomicroscope have been  achived. The 
systematic image errors have reached 50um and 
have not been stable. In the case of the stereo- 
microscope the zoom seams not to be very stable, 
in the case of the Rolleimetric the mounting of 
the different optical elements cannot be repeated 
accurate enough. 
    
The attributes of the optical systems used for 
the CCD-cameras are different. The systematic 
image errors were below 0.3 pixels and not 
significant. It is necessary to use only cameras, 
switched on since a longer time to avoid the well 
known drift of the CCD-values during warm up. 
The examination of zoom lenses, the Vivitar with 
a C-mount for CCD-cameras and the HFT Rolleinar 
for the Rolleimetric, have demonstrated the 
problems of zoom systems. The reproduction of 
the inner orientation based on marked setting 
points was not possible within the aspired 
accuracy. A selfcalibration for every new setting 
is required. Also the strong dependency of the 
radial symmetric lens distortion from the focal 
length has to be investigated every time. 
4. Geometric conditions during 
practical applications 
As result of the above mentioned investigations, 
the different systems were used also for practical 
applications with a threedimensional control 
field like shown in figure 3. Without such a test 
field the camera orientation cannot be determined 
accurate enough. The traditional orientation 
procedure with relative and absolute orientation 
is not possible. The correlations of the elements 
of the relative orientations are listed with the 
values 1.000. But the the bundle solution can be 
used without problems. 
The principal point and the calibrated focal length 
were fixed and the bundle orientation has been 
donewith selfcalibration by additional parameters. 
Inthe case of the use of photos, analytical plotters 
are used for the model handling. The orientation 
parameters of the bundle solution have been 
used also for the orientation of the analytical 
plotters. In addition an on-line correction of the 
fiducial mark position based on the systematic 
image errors was used. 
In the case of on-line photogrammetry with 
CCD-cameras, the exterior orientation was fixed. 
With such a configuration it is not necessary to 
determine every orientation element and a 
two-dimensional control field is sufficiant 
(Jacobsen et al 1990), (Husen et al 1992). 
5. Applications 
The trend in close range photogrammetry is going 
to on-line applications. The major disadvantage 
of this is the limited resolution and accuracy in 
relation to the image size (see table 1). The 
procedure has to respect these limitations. A 
typical example of on-line application of the 
macro photogrammetry is the determination of 
profiles of a toothed wheel. With 2 CCD-cameras 
in a convergent solution the object has been 
imaged. Based on correlation the profiles were