5 -
did not succed to introduce these tie conditions. So, the
experiments with real image data are in complete with this
respect. Nevertheless, I think it is quite interesting to regard
the results we have achieved so far.
|, EXPERIMENTAL RESULTS WITHOUT INTRODUCING 9 K
sT* Üsgp^ Kar
Two different series of experiments have been carried out. The
first deals with real image data without use of readings of sm,
Warm and Kam» @S already mentioned. The second series regards
the numerical stability and correlation values of the system
(chap. 5). For these investigations the constant terms in the
equations do not play a part, so we may treat the complete
system of equations in this respect.
Now, at first let us compare the results we have achieved with
all informations available (see /1/, tab. 3 and Abb. 6 for the
residuals) to those with less informations, that means less
photo series and / or less terrestrial points in testfield 2
and 3, For the comparison we use the Hotelling-T^-statistios,
see /1/, equations (23) - (25) and table 5.
TESTFIELD 1:
Without any risk only one series of photos may be used. This
obviously is a consequence of the size and very high number of
points in testfield 1.
TESTFIELD 2:
You may achieve safe and reliable results if more than one
series of photos are used. But a certain unsteadiness can be
observed.
TESTFIELD 3:
Due to the geometric uncompleteness of the mathematical model
testfield 3 cannot deliver reliable results. For this type of
testfield you cannot despense with the readings of the orien-
tation angles Pam» n» and Kam» as has been possible in case
of testfield 2 and especially of testfield 1.