a) The first set of data consists of the manual measurements of all 
signalized points (tie-points, control- and check-points).It 
corresponds to version b) of the Schnürpflingen test. 
b) For the second set of data natural tie-points were used, the trans- 
fer of which was executed by digital image correlation. It 
corresponds to version c) of the Schniirpflingen test. All further 
details are the same as described in section 2.2. 
In both versions altogether 452 image tie-points were measured, plus 52 
image points of tying control points. As the latter were signalized, 
their image coordinates were taken over from set a) into set b), .as 
described in section 2.2. The total number of 504 image tie-points (18 
per photograph) represent 144 terrain points. 
The internal precision of the digital image correlation, as assessed by 
the least squares correlation algorithm, is again very high. The r.m.s. 
value of all standard deviations of x and y transfer parallaxes is 1.1 
um. The cluster diagram of fig. 2 shows that only 6 out of 326 digital 
image correlations had standard deviations of x-or y-parallaxes » 2 um. 
One correlation at point no. 1662027 could be considered as an outlier. 
In this case the shadow of a tree disturbed the marginal region of the 
digitized array. However, the point was not discarded, as small resi- 
dual errors in the block adjustment did not give reason to. 
o (um) À 
3.047 . 1662027 
Fig. 2 Test Appenweier 
| : : Internal standard errors 
2.0 : of digital image correlation 
at natural tie-points 
  
3.3 Both sets of measured image coordinates were processed in the same 
way through bundle block adjustment with 12 block-invariant additional 
parameters (program PAT-B). None of the observations had to be dis- 
carded as outlier. 
The results of the block-adjustments are summarized in table 3. It 
shows, for version a) and b), the co Values of the adjustments and the 
r.m.s. differences (py) against the 23 planimetric check points and the 
10 vertical check points. The differences contain the errors of the 
geodetic coordinates of check points (estimated to o = 1.5 cm in X 
and z which is equivalent to g = 1.9 um in photo scale). 
- ‘Qu