The purpose of this work is to show how it is possible to transfer to 
analytical photogrammetry, the FOD problem (i.e., the choice of the 
optimal number and kind of points to be considered in photogrammetric 
models) and the SOD problem (i.e., the search of the optimal distribut- 
ion of weights of each pair of collinearity equations relating to the 
plate coordinates read by the plotter). 
The used method consists in choosing a target value for a function of 
the variance-covariance matrix of the unknowns and in a repeated solut- 
ion of the SOD problem, cancelling each time the observations with a 
non significant or non-significative weight. If, as it is the case, the 
target function is related to the precision of the estimate of the un- 
knowns, and the target value of the function is less than the resulting 
value obtained choosing unitary weights, the resulting weights of the 
observations permit a better precision with about the same number of 
observations. The cost of the additional precision is limited, because 
it is only due to the additional work made in laboratory to read the 
plate coordinates and not to an additional field work. 
2. THE BASIC EQUATION FOR THE SOD PROBLEM 
Let the regular Gauss-Markov model be considered for the bundle method 
of block adjustment 
y - u = E(y) = Ax o(A) = nxm, rk A=m 
D(y) = D(u) = 2 = ol p^! positive definite 
leading to the regular system of normal equations 
^ 
Nx - b N: s A'PA , rk Nsm, b: = A'Py (1) 
where A : is the (nxm) coefficient matrix of the increments of 
the unknown ground coordinates and of the orientation 
parameters(3 indipendent elements of the rotation 
matrix and 3 coordinates of the projection centres) 
y : is the (nx1) constant vector given as difference 
between the vectors of image coordinates and their 
approximated values 
u : is the (nx1) vector of residuals to the image coordi- 
nates 
X : is the (mx1) vector of increments of the ground coordi- 
nates and of the orientation parameters 
- 149 -