12 
GIUSEPPE I N GH ILL E RI 
Work on the plotter. 
As we said, once the instrumental coordinates As, £ s and °f the 
taking points are known, the strip is executed according to the procedure of inde 
pendent pairs. 
We place the first photogram in the left camera and the second in the right 
camera ; a relative orientation with <p 1} <p 2 , x x , x 2 and Oj (or w 2 ) is performed ; 
then, we read the coordinates of the pass points. In the experiments we executed, 
the pass points on the model were three : a nadiral point N, a point A and a point 
B. In this way we have at disposal 4 points for the orientation of each model. 
However, it is possible to introduce more than 3 pass points, especially if the com 
putations are carried out with electronic computer. 
Once the operations on the first model are accomplished we put the second 
photogram in the left camera, the third in the right camera, we repeat the above 
mentioned operations and so on for all the models of the strip. 
Of course, besides the pass points coordinates, also the control points coordina 
tes are read. The time required is, practically, the same as in the usual analogic 
connection. 
In fact, comparing the usual connection with the semianalytical one, we can 
observe that no difference of time comes from the use of the relative orientation, 
asymmetric for the first and symmetric for the second ; in the analogical method 
it is necessary to scale the model, namely to determine b x , but this operation 
does not influence remarkably the work time. 
Concerning the control of the models, it is necessary to make a comparison 
among the different types of aerial triangulation. In the analytical triangulation 
the model is not formed by the operator and, therefore, there is no immediate 
control; in the analogical method we can control the model and also the discrepancies 
in height on the pass points ; in the semianalytical procedure, the formation of the 
model is controlled through the observation of the residual parallaxes, but no 
control is possible, of course, in the heights of the pass points. 
Present computation programs and development of Semianalytical 
Triangulation. 
The computation program has been drawn up, at first, for the Univac USS 
90 and, afterwards, for the IBM 7040 in Fortran. We do not give here any detail, 
since the program, also with a complete automation, is quite easy. However, we 
can say, to show the small number of computations, that the time for the compu 
tation of a 10 photograms strip is about 1 minute on the IBM 7040. 
The method of electronic computation we followed includes, of course, the use 
of the least square method, since we used a redundant number of orientation points 
and other computation techniques that could be simplified in the case that the 
calculation is performed with a desk machine. 
A schema of desk machine computation has not yet been established, but,