We meetthese difficultiesiwe know the orientation parameters, but we 
do not know the terms XS,» YS; ° 52d, XA, Y 9: XC, which define the coordina 
tes of the ‘perspective center’ and the matrix R of the point considered. In 
order to know these it is necessary to know Y, which, in the hypothesis of 
an adjusted camera, coincides with the coordinate f. On the overall we 
should already know the internal direction tangent t which on the contrary 
is one of the unknowns sought. y 
This consideration shows us the practioal difficulty in solving the 
problem, but also indicates to us the way it can be solved. 
By solving equations 3) with respect to 2 i we obtain: 
* 
  
  
cdi e doe dS 
x |) tx t Y 1]. m i. by 1 m. e ig 1] = | r, 
‘ [og tg tlm [y ty 1] eom QUE 
7 |r teddy [ox bg Hho By 0g By 1 em |r 
however, the preceding formulas are a false expression of the internal di 
rection tangents as a function of orientation parameters, in so far as 
both the terms ty, ty and the matrix r, contain the parameter {= = arc 
tg (-t ). It will therefore be necessary to compute t and t in a way a 
nalogolis to what is done in the case of the computation of the orientation 
parameterss that is we must use an iterative computation process. 
The space resection computation performed on the theoretic model 
showed that the formulation of the space resection problem is sound, more 
over tests carried out confirmed what said in the foregoing points a) and 
b). In fact, the computations showed that the mean square error by which 
the unknown o was determined, was much higher than that of the other un 
knowns, Moreover, the analysis of the variance-covariance matrix showed 
that it was impossible to determine both the unknowns v, and 9. Therefore. 
we can conclude that it is possible, by means of the here described com 
putation, to determine 10 of the 12 unknowns, while the parameters 9 and 
V must be determined in a different way. That is to say that it is possi 
ble to determine the component S only, but not the components 3S and S3 
of vector S (see 3.2). = y Z 
The other unknowns are determined with the same accuracy that it is 
possible to reach in the field of photogrammetry with plane cameras; of 
course the internal orientation parameters must be known with the required 
accuracy and the adjustment conditions must be verified. 
4 - THE RELATIVE ORIENTATION OF 2 PANORAMIC PHOTOGRAMS, 
  
4.1 - GENERAL OUTLINE FOR THE COMPUTATION OF RELATIVE ORIENTATION. 
A careful study brought to the conclusion that the problem of relati 
  
   
   
    
  
  
   
   
    
   
  
  
    
   
   
  
  
  
  
  
  
  
  
  
    
  
  
  
  
  
  
  
  
  
  
    
ve ori 
the fo: 
2) 
Fay 
and th 
the se 
tions . 
sent Oo 
T from 
tive Oo 
to the 
(p) 
X-X 
1 
2.758) 
Ts 
(p) 
m 
pint) 
(1) 
where: