6.3. 
6.3.1. 
— 15 — 
a) Executing agency:the estimator o0 1 varied by as much as a factor 2.7 
  
in the results of a single block measured by different organisations. 
This result is even more significant bearing in mind that the photography 
was taken with a plate camera, oo 1 being thus anyhow extremely small! 
b) Type of camera: in blocks of roughly the same size, Ÿ 100 models each, with 
photo scales varying between 1:5000 and  1:8000 and measured by the same 
organisation, o 1 was found to vary between the limits 2.3 — 2.5 4 for 
plate cameras and 2.9 — 3.7 u for film cameras. The mean values for the 
plate and film cameras varied by & factor 1.3, illustrating that the camera 
type is à significant influencing Factor. 
c) Restitution instrument, o]! varied by a factor 1.6 in a block of 205 models 
  
measured by the same organisation but on two different stereocomparators. 
No reliable conclusions can, as yet, be drawn concerning the influence of 
other parameters, such as photo scale, number of orientation points, etc. 
on the estimator o 1, and in addition, more conclusive proof is also 
desirable for the influence of the three parameters mentioned above. 
However, the values of o! found so far, (90% less than | u ), are se 
small compared to the accuracy of image coordinates known from experience, 
and this thus poses the question as to the reliability of the estimator a 1 
to describe the image accuracy. Theoretical studies (Kubik  [10]) have since 
shown that o 1 is incapable of measuring the occurrence of systematic 
image deformations and we may thus conclude that a modified estimator will 
have to be used to define the image accuracy. 
The estimator o2 
  
This second accuracy measure in the triangulation process, (see fis. 3), is 
obtained from the computation of the standard deviation of unit weight after 
a least squares strip or block adjustment. Since this estimator has a different 
physical meaning for different types of adjustment, the estimator o 2 will be 
discussed separately for the different computational units used. 
Plates as computational unit (bundle adjustment) 
The estimator o 2 determined after a bundle adjustment is a measure of the 
accuracy of points in the photographic image. 
The sample of 35 estimates and varied 
2-11 u 
o 2 (bundle) had a mean value of 6 u 
in magnitude between , 90% of them being less than 7 u 
In order to investigate the influence of various parameters on the estimator 
5 2 (bundle), use was made of a sample of 24 blocks flown with the same camera 
N 
\ 
6.3.2 
e e 
© 
and me 
differ 
bridei 
The re: 
scale 
negati 
distanc 
of any 
small s 
even wt 
(bundle 
which, 
be muck 
Models 
The est 
or sect 
model Tr 
The san 
value c 
The der 
followi 
a) Brid 
ment 
“brid