GIUSEPPE INGHILLERI 
l6 
Some results. 
The experiments have been performed on strips 2-6-1 and 2-6-4 of the Rei- 
chenbach Polygon (O.E.E.P.E. Comm. B). Both the strips have been taken with 
the Wild RC 7 camera, p = 100 mm ; size 15 x 15, at a relative altitude of 1 200 m. 
Strip 2.6.1 has 10 models and strip 2.6.4 has 9 models. Strip 2.6.1 has been 
measured 3 times and strip 2.6.4 2 times. 
In the following tables we show two examples of the results of the linear and 
parabolic transformation. 
For the linear transformation four control points at the end of the strips 
have been used ; for the parabolic transformation nine control points have been 
used : three at the beginning, three in the middle, and three at the end of the strip. 
Some considerations must be done principally for the results obtained in the 
connection of the models. Since the absolute orientations have been performed 
with 4 points, we may, in practice, give a meaning to the discrepancies in the 
common points coordinates. 
The mean square values of these discrepancies deduced from 5 strips are the 
following (in gm at the scale of the plate) : 
mx 
my 
mz 
2.6.1 
I 
6 
23 
29 
2.6.1 
II 
9 
32 
36 
2.6.1 
III 
7 
20 
23 
2.6.4 
I 
6 
l6 
17 
2.6.4 
II 
7 
29 
23 
These values show the efficiency of the connections between the models. In fact, 
before the starting of the experiments, we feared that the connections of the models 
should not be very satisfying. Other confirmations of the good connection can 
be found in the tables. We can observe that several points have two vectors that 
represent the planimetric error, and two vectors showing the altimetric error. 
This is the case of those control points that have been surveyed in two consecutive 
models. The examination of these points shows the accordance, generally good, 
between the corresponding errors. 
A few words must be said concerning the influence of the errors on the coordi 
nates of the taking points x a , y B , z s and x&, yd, Za. 
We know that these coordinates can be obtained in the Stereosimplex mod. 
Ill, with a remarkable precision. However, we have introduced some errors in the 
/, y, z of these points and we have noticed (as the theory itself shows) that these 
errors cause systematic errors that can be perfectly eliminated with a parabolic