2) 
ter 
by 
3) 
4) 
on 
15) 
16) 
Vs 
7) 
18) 
en 
the 
ral 
en 
jon 
ind 
ith 
the 
to 
Dg 
9) 
need not measure the approximations of angular 
orientations, even more, the method is available to the 
block photographied with amateur cameras. 
4) The parameters are more tightly linked in this method 
than in bundle one, and many of the parameters are 
directly observed, so the condition of normal equatons is 
much better than that in bundle one. 
4. THE STRUCTURE OF NORMAL EQUATIONS 
The relations among parameters are strongthened in crown 
adjustment, though the structure of normal equations is 
changed too. We all know that a favorable structure of 
normal equations can spare much time and space of 
computer. 
  
Fig. 7 
Now, let's take the block shown in fig.7 as an example 
to discuss the structure of normal equations by crown 
method. The photos and ground points are numbered in a 
direction perpendicular to that of the strip. The photo 
numbers adopted here are the same as those of the ground 
points near the principle points of the photos. Fig.8 is the 
structure of the observation equations. From fig.8 we can 
clearly see that, it is only base points that interferes the 
structure of normal equations, so we can teat them 
carefully to get a favorable structure. The following 
principles are helpfull to deal with the problem: 
1) Select base points as few as possible, in other word, 
several photos can share the same base points. In our 
example, all groung points may be as base points, though 
we select just 4 points(number 31, 32, 34 and 35). 
2) Select geodetic points as base points whenever 
possible, because the geodetic points will not be reduced in 
a reduced normal equations. 
3) Remain the base points in reduced normal equations. 
Fig.9 is the structure of normal equations for the block in 
fig.7 by following the above principles. It's clear that 
except the base point terms, the structure is the same as 
that of bundle method. 
  
Fig.10 Fig.11 
  
   
   
   
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
u 
12 
1/18 
Mu 
15 
21 
2 
12/2 
2 
25 
a. = 
3|33 
le m 
3 "b E 
21 
2 
2/13 
2 
BE — 
2 "A E 
IPs — m 
Hu ^ m uuum 
% a = 
s ; 
42 
4|43 
4A 
45 
2 > E 
3|3 "E 
s -— m 
3 m = 
[ [4| 
42 
Il4|« 
44 
[145 
a 
5|5 
5 
5 
known " ^ = 
points ^ 
Fig. 8 
Fig.9 
Fig.10 and 11 are the structure of the reduced normal 
equations in which the photo elements are reduced. In fig. 
10, the base points are place at the end of the equation 
forming a margin of the matrix. This is suitable to the 
solution of banded banded matrix. In fig.11, the base 
points are placed just after the parameters which are 
related to them. Obviously, there is no margin now, and 
the profile of matrix is less than that of fig.10, but the 
matrix is not the equal-width banded one. This is suitable 
to the solusion of inequal-width banded matrix.