Full text: Photogrammetry for industry

2) 
b 
b 
29. 
v 
‘ 
ept 
"d 
got 
à d 
for 
the global test - taken from Baarda's nomograms in /3/5-"ig varying from 0.22 un- 
til 0.45. 
Table 4: Global test of the null-hypothesis Ho! E (85) = st for all network 
arrangements and gross error cases. 
  
  
  
  
  
  
  
  
  
Arrange- o for gross-error cases 0 
ment | | | c(e) a without 
2 | b | € 54 d gross erry. 
A 2.06 | 2561 [1.40 | 1.75 |:1:18 | 0.272 1.40 
B 0.92 | 1.07 | 0.57 | 0.99 | 1,18] 0.22 0:57 
C 1.30 / 1.39 | 1,13 1.23 | 1.04} 0.37 0.96 
D 1.42 | 1.58 1.28 | 1.34 | 1.02 0.45 1.21 
E 0,85 | 0.98 | 0.83 | 0.94 | 1.02] 0.45 0.72 
42 
0 = m , tے(9) from F(1-o,r,*) 
O 
Regarding the results of Table 4 reasonable doubt is put forward with respect to 
the practical application of the global test. In the gross error cases both ver- 
sions B and E don't lead to the correct statement, i.e. to the rejection of the 
null-hypothesis. Though there are fairly good presumptions, because the expecta- 
tion E (07) = 0° is a-priori known by the utilization of a random generator even 
under the null-hypothesis (when no gross errors are included) the critical values 
are exceeded twice. Of course the random generator produced discrete random 
values, whose standard deviation differs more ore less from the a-priori chosen 
value of 5 um, but this problem is even intensified in real practical] projects: 
What is then the expectation of 527 That is probably the reason why Pope /17/ 
doesn't use the global test - generally its power of assertion is not sufficient. 
Hence it is not recommended in practice to make the setup of individual tests 
(data-snooping) dependent on the result of the global test. The procedure of data- 
snooping should be applied in any case if reasonable suspicion exists that gress 
errors may appear, what is the standard situation in practical projects. 
In this way we continue with the individual checking of the residuals. To avoid 
at least partly the disturbing effects of correlations between the residuals the 
sig egy consists in rejecting those observations, for which the test criterion 
Vii 
Wi "gore is maximal. 
v. 
3 
As a result it was found out that all a-priori known gross errors have been de- 
tected, expect in the following cases: 
]. Gross error case c in arrangement A and B: zero variance situation 
2. Failure with the gross error case a of arrangement B: 
the x'-observation of point no. 3 at image 3 led to max (wy) and 
thus to the prior rejection, although not including a gross error 
3. Failure with the gross error case b of arrangement B: 
the x'-observation of point no. 9 at image 3 led to max (wy), 
although not including a gross error 
  
 
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.