= 
7 
S. 
1 
- 
i 
> 
1 
(95 
ms 
pe Im >< 
t- 
vont rre 
"s ll 
D 
—— X 9X eX 
1 
J 
k 
] 
1 
J 
RO 
1 
Hs 
ld 
es 
The non linear conditions (7) have to be ‘linearized "with 
respect “to X, Y."Z of points i, 3, 'k for tne line. The same 
has to be done with condition (8) for tne plane (linearized 
with respect to points i, ij, k, I). Every additional point on 
the line gives raise Lo ons mors pair of line-conditions (7), 
every additional point on the plane gives raise to one more 
condition (8). 
Remark: 
Notice that observations and conditions are included in 
two different ways in the adjustment. First image coordinate 
measurements of every object point lead to equations (1). 
Secondly observations and conditions itself are included by 
linearized equations (2), (2), (4), (5), (65), (70, (8). For 
conditions a relatively high observation weight is chosen to 
fulfill the condition. 
2.3 Solution of adjustment: 
For the solution of the least squares" adjustment the 
method of conjugate gradients is used. This method for solving 
symmetric definite equation systems has been developed by 
Hestens and Stiefel (1952). Applications in geodesy has been 
demonstrated by H. BR, Schwarz (1970), Gruendig (1930) and 
teidler (1980). 
This iteration "procedure is directly working with the 
observation equations and therefore saves the memory- and time 
consuming computation and solution of the normal equations. A 
special quality of the gonjugate gradients method is the quick 
convergence in a local area. 
Besides the simple and quick computation and the limited 
need for core memory, another advantage occurs, which is due to 
all gradients methods. For adjustment of observation equations 
With rank deficiency (for example free networks, self-contained 
partial blocks of images) a transformation onto the approximate 
coordinates 1s automatically executed. For a free adjustment 
of the network’ this approach offers a plausible solution 
without additional expense. 
A short overview of the method of con 
described below. 
Given are the observation 
(D 
C 
í- 
09 
to 
ct 
(b 
09 
3 
£0 
co 
+ 4 
(b 
Ij 
Ct 
u 
+ 4 
U 
Ca. 
ct 
3 
uations 
. 
(D 
2 
NO 
A 
As = 1 = y ( 
1 à » + - * 
leadin unotion 
This is 
     
09 
ct 
O 
£D