near hypothesis (29).
Here a bivariate test criterion is derived, which can be applied if two additio-
nal parameters are highly correlating.
| The null-hypothesis is formulated as
|
|
hhh”
Ho ]'dz = d2° (42)
2.2:2,1 852.4
á Suppose dz' = O and dz! = (dz, dz, ), then T results in
1 q 9.2.31 Lat
A A 23123 Zi ;
Fou (A 4 1 UE ^: (43)
o qd7Kk21i Qzkzk dz, . {
The application of various test criterions and a possible strategy for additional
| parameter rejection was demonstrated in Grün /10/ with the use of a practical
aerial triangulation example. Because of the extension ofaerial triangulation í
| systems one has to work with some neglectings concerning the application of com-
pletely rigid statistical tests; thus the concept of a-posteriori orthogonaliza-
tion of the whole system is hardly to realize. However, close-range systems opena
wide and interesting field of application of sophisticated statistical tests. So
this subejct should be a main objective for further studies.
2. Significance testing of residuals at check points and control points
It is a widespread and useful procedure in empirical accuracy studies to compare
the set of coordinates of a photogrammetric adjustment with their "true" values,
mostly obtained by more precise observation methods. Then the r.m.s. values of
the residuals serve as absolute accuracy measures to check the photogrammetri -
cally achieved accuracy. é
If we denote the estimated residual vectors at check points as
^T ^ ^
AX z (AX 9 SAX ) E
1 PL : ;
^T A ^ PL = number of planimetric
AY 5x (AY, > >AYp, ) ; check points (44) v,
oT °° S $ pz = number of height
AN (421, + 38Zp7) , check points
with the components
2] J, 8p
A S '4Ph } ; iL:
M qe m Murus Kk buo ips
AS. z Ys = poe ’ &Ph, yPh, ZPh = estimated photogramme - (45)
A A tric coordinates
AZ = 7° - 27 S S S
k k k , X> oY iy 2 = "true" coordinates
then the r.m.s. values are defined as
A A
A2 = AX TAX ne i AY TAY ^e A212
T , MN 9 H e TECTUM ,
X iiL que PL b sz (46)
^2 ^2