wt. Jul LW 
3 
Figure 5 shows the test block and the individual control distri- 
butions, used in planimetry and height, The terrestrial points, 
which didn't serve as control points, were used as check points. 
(180 to 250 occasionally), 
- Figure 5 - 
For all 4 strips of the test block the same project parameters 
Can be assumed. From there it follows that only one group of 
additional parameters has to be put up for the whole block. The 
justification of this measure was confirmed by test adjustments 
with strip invariant correction terms, which didn't show signi- 
ficant variations of the parameter amounts from strip to strip. 
Because all correction terms are treated as block invariant para- 
meters the present test is not able to demonstrate the full 
efficiency of the suggested strategy (see 3.3). The separation 
of the block invariant correction terms into significant and in- 
significant terms however, can be shown here. 
The significance tests were performed on the 99 Z level, From 
the beginning it shall be mentioned that the obtained correction 
terms proved as only slightly dependent on the control distri- 
bution. With the poorest density the additional bundle parameters b;, 
b2, bg, b;, bg and bj, were found as significant, Their common 
effect on the image points represents the systematic image de- 
formations and is shown in figure 6 (b = 92 mm). The maximum 
values amount to 9 um in x and in y, 
  
- Figure 6 - 
The further results of the bundle block adjustments are summa- 
rized in table 1. Because all image coordinates were treated as 
observations of weight 1 the standard deviation of unit weight 
oo here directly represents the mean accuracy of the image co- 
ordinates. Without self calibration og is highly dependent on 
the control distribution used, This clearly indicates the 
presence of systematic errors, which influence the residuals 
and og the less the poorer the control density is. Simultaneous 
self calibration reduces og by a factor 1.3 to 1.6 and elimi- 
nates the dependency on the control distribution almost com- 
pletely. With cg = 3.2 um the accuracy of the image coordinates 
is close to the accuracy limit, attainable today at all. 
- Table 1 - 
In table 1 the accuracy of the adjusted block coordinates X» 
and z is represented by the RMS values ux and uz» which were 
determined from the coordinate errors at tle planimetric check 
points and at the check heights. Both, y and u; are reduced 
to the photo scale. Simultaneous self calibration improves the 
accuracy the more, the less favourable the control distribution 
is. The factor of improvement increases up to 3.3 in plani- 
metry and up to 2.3 in height, 
Meere iecit — m m m S - ——" —