The accuracy figures ag, ux,y ànd uz, obtained with self cali- 
bration, are slightly smaller than the corresponding results, 
which Bauer and Müller had published in |8|. These relate to 
a Oberschwaben subblock, sonsisting of the wide angle strips 
1, 3, 5, 7, 9 and were obtained with 3-4 correction terms (3 
successive adjustments with 4 additional parameters each), 
With the block adjustments by independent models the planimetric 
correction terms pi, po, pa» Du, pg and the height terms hos Mas 
hy, hs have proved as significant. Their common effect on the 
model points represents the systematic model deformations and 
is shown in figure 7 (b = 92 mm). The maximum values amount to 
10 um in x, 7 um in y and 11 um in z. As was to be expected the 
results of figure 7 agree well with the model deformations 
being computed from the systematic image errors of figure 6. 
- Figure 7. - 
The individual results of the planimetric block adjustments by 
independent models are summarized in table 2. Because the model 
coordinates x, y were treated with weight 1, the standard devia- 
tion of unit weight cgp directly represents the mean accuracy 
of the planimetric model coordinates. With simultaneous self 
calibration this accuracy figure decreases to gop = 4.3 um. The 
corresponding figure HX,y 2 which estimates the mean accuracy of 
the adjusted block coordinates in x and y, is improved by a 
factor 1.6 t0 2.9, 
- Table 2 - 
Table 3 shows the results of the height block adjustments by 
independent models. cop here represents the standard deviations 
of the model heights and uz describes the mean accuracy of the 
adjusted heights of the block. The accuracy improvement, attained 
by self calibration, is much.smaller than in planimetry. This is 
true for cop as wel] as for ug. The only one exception appears 
with the extreme control distribution i = 25, where the rather 
poor accuracy uz = 65.0 um is reduced to the reasonable value 
uz = 26.7 um. 
- Table 3 - 
The results, listed in table 1 and in tables 2 and 3 were ob- 
tained from the same data material. Therefore they can be used 
for an accuracy comparison between bundle and independent model 
adjustment. Without self calibration most of the bundle results 
ux,y and uz are larger than the corresponding figures of the 
block adjustments by independent models. Obviously this is caused 
by the systematic data errors (see also |3|). 
As soon as the systematic errors are compensated adequately, 
which is guaranteed by simultaneous self calibration, the situa- 
tion changes and the bundle results prove as superior, as ex- 
pected by theory. Table 4 shows the accuracy figures ux,y and uz, 
obtained with both adjustment methods and the accuracy ratios 
Models ^ "bund]es' The maximum ratio is 1.2 in planimetry and 
S00 OT = 
3 3-5 UO =<TO="T X 2 n»3o0(0D Uo VL 
= OO. OM © — =f e 
ct UO X 
pU 
9 
N e+ © = 35060 =F —h