lo 26 3o 
value % value % value % 
  
  
CIC I 0.0182 67 0.0364 88 | 0.0547 99 
  
CLC 2 0.0167 70 0.0335 97 | 0.0503 100 
  
  
  
  
  
  
  
  
  
  
CEC3 0.0198 28 0.0397 67.|. 0.0393 84 
Table 5: Percentage of LME within expected deviation 
The extended model for camera calibration achieves significant 
improvements of accuracy as finally pointed out by results of 
the same data sets with conventional bundle-invariant camera 
model (Table 6). An improvement up to 25um in object space 
could be achieved including image-variant parameters. 
  
  
  
  
  
  
  
  
  
Oo [um] RMS xyz) Relative 
precision 
CLC] 0.553 0.0655 1:41000 
CLCO 0.506 0.0611 1:44000 
CLE3 0.609 0.0747 1:36000 
Table 6: results by conventional camera model 
4. CONCLUSIONS 
Because of nowadays mostly used high-quality cameras — 
partial-metric cameras - new mathematical approaches are 
necessary to model all instabilities and camera based influences 
sufficiently. With the implementation of image-variant principal 
distance, image-variant principal point and an additional 
correction grid to model remaining sensor and lens based 
influences these instabilities can be considered. In general the 
adjustment results, especially the variation in image-variant 
parameters, are highly dependent on imaging direction. The 
variation in y-direction points peaks up to 33um to the mean 
value. Using an invariant interior orientation the effects of 
image-variant part are smeared to other parameters. The 
smearing effects cause probably wrong results. Data sets with 
high instable interior orientation might not be computable by 
conventional approaches. The mentioned data set of the DCS 
Pro Back 645M proofs this assumption. The CCD-array is 
mounted to the camera body and relocatable in x-direction. An 
offset of about 100um during image acquisition with the DCS 
645M was only computable with the described image-variant 
mathematical model. Significant improvements in accuracy 
with respect to a modelling with bundle-invariant interior 
orientation have been achieved. 
The presented bundles do not figure optimal configuration 
because of almost exclusively fulfilling images and influences 
by the image measurements of the object point with small 
ellipses caused by point diameter of only Smm in object space. 
Improvements of image acquisition configuration need to be 
required. The 3-D testfield for verification of the bundle results 
needs closer research. At least one additional measurement of 
the reference points is necessary to eliminate the remaining 
length-dependent part in the length measurement errors. The 
overbalancing part of LME in the range of +40um under 
consideration of instable parts and with careful handling of the 
camera point high-accuracy 3-D results. 
5. FURTHER INVESTIGATIONS 
For further project work a major research will be addressed to 
the improvement of imaging acquisition configuration. The 
optimal number of images and images per imaging direction, 
camera stations by optimal efficiency need to be verified. This 
investigation depends on the 3-D testfield, the reference 
coordinates and the knowledge of the testfield’s behaviour with 
respect to different exterior influences. 
Beyond analyses concerning accuracy of three-dimensional 
objects and verification based on the German guideline 
VDI/VDE 2634 will be done. 
6. REFERENCES 
Dold, J. (1997): Ein hybrides photogrammetrisches 
IndustriemeBsystem höchster Genauigkeit und seine 
Überprüfung; Schriftenreihe Universität der Bundeswehr 
München, Heft 54. 
Fraser, C.S., Shortis, M.R. (1992): Variation of Distortion 
within the Photographic Field; PE&RS, Vol. 58, No. 6, June 
1992, pp. 851-855. 
Jantos, R., Luhmann, Th., Peipe, J., Schneider, C-T. (2002): 
Photogrammetric Performance Evaluation of the Kodak DCS 
Pro Back; ISPRS Symposium Commission V, Corfu 2002. 
Kraus, K. (2000): Photogrammetrie, Band 3, Topographische 
Informationssysteme; Dümmler Verlag, Bonn, p. 188ff. 
Luhmann, Th. (2000): Nahbereichsphotogrammetrie; 
Wichmann Verlag, Heidelberg. 
Luhmann, Th., Wendt, K. (2000): Recommendations for an 
acceptance and verification test of optical 3D measurement 
systems; International Archives for Photogrammetry and 
Remote Sensing, Vol. 33/5, pp. 493-499; Amsterdam. 
Maas, H.-G. (1998): Ein Ansatz zur Selbstkalibrierung von 
Kameras mit instabiler innerer Orientierung; Publikationen der 
DGPF, Band 7, München 1998. 
Munji, R.A.H. (1986): Self-calibration using the finite element 
approach; Photogrammetric Engineering and Remote Sensing, 
Vol. 52, No. 3, March 1986, pp. 411-418. 
Munji, R.A.H. (1986): Calibration of non-metric cameras using 
the finite element method; Photogrammetric Engineering and 
Remote Sensing, Vol. 52, No. 8, August 1986, pp. 1201-1205. 
Shortis, M.R., Robson, S., Beyer, H.A.(1998): Principal point 
behaviour and calibration parameter models for Kodak DCS 
cameras; Photogrammetric Record, 16(92): 165-186 
Tecklenburg, W.; Luhmann, Th.; Hastedt, H. (2000): Camera 
modelling with image-variant parameters and Finite Elements; 
Optical 3D; Vienna 
VDI/VDE 2634 (2001): Optical 3-D measuring systems — 
Imaging systems with point-by-point probing. VDI, Düsseldorf. 
239... 
as