c) Together with the measurements b), without changing the interior 
orientation of the plates in the Planicomp, natural tie points were 
selected close to the previously measured signalized tie points. The 
local image areas of the natural tie points were digitized by the 
CCD video cameras, externally stored and later processed for digital 
image correlation, as described in section 1.3. 
d) In additon to measurements c) another set of data was produced with 
the Planicomp by digitizing the signalized tie points with their 
surrounding areas and using them for point transfer by digital image 
correlation. This set of data is not of direct practical importance. 
But the opportunity was used to compare the performance of digital 
image correlation of natural and of signalized points. 
The image coordinates for the cases c) and d) were obtained by digital 
image correlation of the externally stored digital arrays. The internal 
precision of the digital correlation (not to be mistaken for precision 
of the transfer points) is assessed by the least squares correlation 
algorithm for each individual correlation. The average standard errors 
of x- and y-parallaxes at natural tie points are 1.2 um.Fig.1.1 and 1.2 
show the cluster diagrams of the individual standard errors. The values 
are smaller than 3 um (2 um) for 97 % (80 %) of the natural tie points 
and for 90 % (61 %) of the signalized tie points, respectively. 
Test Schnürpflingen, internal standard errors of digital image correlation 
Fig. 1.1 natural tie-points Fig. 1.2 signalized tie-points 
gy Cum) 
4.0 ay lum) | 
6.0 
3.0 5.0 
4.0 
    
  
  
  
m 
T T T T 
1.0 2.0 3.0 4.0 c, (um) 1.0 2.0 3.0 4.0 5.0 6.0 a, (um) 
During the subsequent block adjustment 20 digital correlations with 
natural tie points had to be discarded because of excessive residuals. 
15 of them showed poor correlation and could have been immediately and 
automatically identified as not acceptable if appropriate threshold 
values would have been applied. In order to maintain a strict basis for 
comparison those discarded tie points were also discarded in all other 
data sets, leaving 405 image tie points. Thus, the 4 cases have identi- 
cal geometric configuration. 
The signalized control and check points were not measured digitally in 
the data sets c) and d). They were taken over from measurements b). In 
this way the comparison gives directly the effects of digital point 
transfer. A condition is, however, that the digital arrays are precise- 
- 21 -