graphy 
alll 
ach, with 
> same 
for 
^ ihe 
> Camera 
> models 
ators. 
1ce of 
, etc. 
30 
s 
~ience, 
Mor a 
ave since 
LC 
Yr will 
18 
fier 
.fferent 
11 be 
the 
| varied 
ia tor 
> camera 
6.3.2 
zZ 1Ÿ = 
and measured by the same organisation on the same stereocomparator, but with 
different photo scales (between 1/4.000 and 1/41.000) and with slightly different 
bridging distances (perimeter control between 1 and 4 models). 
The results indicate that. 4.2 (bundle) is directly proportional to the photo 
scale (1.0. independent of photo scale when o¢ 2 is reduced to microns in the 
negative scale) and that it is independent of slight changes in the bridging 
distance. o 2 (bundle) is too small to detect the influence 
The sample size of 
of any other parameters and this may also affect the reliability of the extremely 
small values found for this estimator, namely 90% of the value less than TAE 3 
even when using film pholography. As wiih 6G 1, it is-also suspected that g 2 
(bundle) is incapable of taking full account of systematic image deformations, 
which, for example, from reseau measurements on film photography, are known to 
be much larger. 
Models or sections as computational unit 
The estimator o 2 determined 
after a least squares adjustment with models 
or sections as computational unit is a measure of the coordinate accuracy of 
model points before adjustment. 
The sample of 112 estimators o 2 (model) varied between 4-40 wu 
25H 
, with a mean 
value of ) 90% of the estimates being smaller than 
15 u 
The dependency of o 2 (model) on various parameters was now investigated and the 
following conclusions arrived at: 
a) Bridging distance 
  
fig. 4 illustrates the results of repeating the adjust- 
ment of 2 blocks of respectively 200 and 25 models each, varying only the 
bridging distance. 
    
  
   
  
  
02 Hu 
$ 
164 
14- 
_— 200 models 
12} 
10+ 25 models 
T T T T T > d 
4 8 12 16 20 models 
fig. 4 Dependency of 0 2 (model) on bridging distance