dual y-parallaxes observed on the control points after the absolute orien-
tation of the photogrammetric models, the absolute values of the mean
square errors concerning the accuracy of the photogrammetric determi-
nation of check points.
A summaty of all these values is given, test by test, in the Table I here
enclosed.
The first research carried out on these data had the purpose of establishing
their statistical parameters.
Aiming to analyse the distribution of the frequency of each character
summarized in the said table, the arithmetic mean of the summarized
obsetvations, the variance, the standard deviation, the standard errors of
the means, the standard errors of the standard deviation have been
computed. The results of the computations have been shown in Table II
here enclosed.
Having concluded this first research, we have proceeded to determine
the degree of linear dependence or covariation between the more signifi-
cant variables of the problem under consideration, that is to say: the
errors corresponding to the alterations introduced on ground control
points the mean square errors of the residual y-parallaxes observed on
the control points for each photogrammetric model and the errors due
to the photogrammetric plotting of the said control points obtained from
the differences between actual topographic coordinates and machine coor-
dinates transformed in the ground system.
Of course the measurement of the dependence degree has been obtained
computing the linear correlation coefficient corresponding to the diffe-
rent pairs of the above said variables taking into account the totality of
each considered error, model by model. In fact we have deemed advisable
for the undertaken research, to conglobate the two vertical and horizontal
components of error.
Showing the above said variables by zi, 7e, vs respectively, corresponding
to the values summarized in columns 4, 5 and 8 of Table I, we have
used for this purpose the formula:
n /n , n \
n Xx,.y, (=x) (2x)
i=1 fue] i=1
) r=—p
n n \ 2
CQ 9 4
aXx—(x x |
n n 2
nXyi (Xy.
i=1 i=l, i
i 1 (i=l
