Il 
11. 
each point in areas la, lb, and II respectively, are: 
m 
£ èik + = 0 (a) 
q = 1 
11 k VI llm 
... 2.18 
Y_ (jL r ¿Lur + JL r ) = 0 
iwn r !UV 1K " - r 
r = 1 
uu 
1 .è 
ik 
= o 
(t) 
(c) 
UxilP tur 
P = 1 
The contribution of each individual point may be processed separately and 
added to the final form of the normal equations 2.13 in a step by step 
procedure. 
Solution for the corrections, A^ is an iterative procedure, terminating 
when the corrections reach a pre-selected value- At this point, output 
from relative orientation consists of the most probable va]Ues for the 11 
unknown parameters (Y . , Z . , oj. , , Z . ). 
Using the most probable values for parameters and the final corrections, 
—ik 3 P^- a ^ e coordinate measurement residuals are determined by least 
squares adjustment using equations 2.l6 and 2.17 • The diagonal nature of 
3, 12 
the Q matrix allows computation of residuals for each point individually 
Thus, formation of normal equations and determination of residuals are 
repetitive processes, independent of the number of points carried in the 
triangulation and ideally suited for use on electronic computers. The 
maximum size matrices which must be inverted are: (a) 3 x 3 to process 
points for normal equations and solve for residuals; and (b) 11 x 11 to solve 
for corrections, A., using equation 2.15 • The standard error of unit weight 
of the adjustment, a, is evaluated using 
a = 
T 
V P V 
... 2.19 
f ~ g 
in which f = total number of independent condition equations, 
g = total number of unknowns, 
and V T P V = the weighted sum of squares for residuals of all triangulated 
points.