281 
object 
fig 6 
configuration for 
video cameras 
6. Data handling 
The images taken by the stereo microscope, the Rolleiflex metric 6006 and 3003 
and for checking purposes also displayed digital images from video cameras are 
handled in the Planicomp and the DSR 11. The photo orientation by bundle 
adjustment is computed by the Hannover program system BLUH in the case of more 
than 2 images or by program BUNOR. The 2-photo-bundle orientation program 
BUNOR is managing the whole process, including computation of approximate values, 
error detection by data snooping and selfcalibration by additional parameters within 
few seconds. 
The standard programs for data aquisition can be used based on transformed photo 
orientations. For sufficiant accuracy the systematic image errors, determined by 
calibration and self calibration, must be respected by a correction grid, which will 
give a correction to the position of the floating mark based on bilinear interpolation 
(Picht 1987). This flexible possibility for correction of the photo geometry allows also 
the handling of photos with large differences to perspective images. 
The data aquisition with the digital plotter, developed by the Institute for 
Photogrammetry of the University of Hannover (Lohmann et al 1989), includes the 
advantage of optimal contrast enhancement and a more simple support by 
correlation. 
A manual height measurement in a model taken with 6-fold enlargement by the 
3003 and a base to height ratio of 1 : 2.6 has lead to an accuracy of the height 
determination of metalic surfaces of +/- 3 pm in the object space, corresponding to 
+/- 7 pm standard deviation in px. With automatic correlation based on the same 
photos, digitized by an Optronics P1700, in the digital plotter an accuracy Sz = 
+/- 5.8 pm was reached. Images of the stereo microscope only have enabled Sz = 
+/- 13 pm. 
The digital plotter has the advantage of a human interaction in the case of failed 
correlation. A small percentage of missed points is not important if only 
characteristic parameters of surfaces, like roughness, have to be determined. This 
can be done close to on-line with video cameras attached to a PC. Such a system 
is under development.