be 
el 
1t 
n 
  
  
relative and succesive orientations are executed 
by machine or computation. In both cases the 
absolute orientation of the first model, as has 
generally been done in usual mechanical method, 
is of no need. Only the condition must be fulfilled 
that the sufficient number of control points will 
have some adequate distribution in the whole 
strip. Then the absolute orientation will be done 
as was expressed in article (2-1) by computations 
for a whole strip at once. 
A result of an example applied for the case is 
as follows—a representative one of about 2,000 
models practised during this one year in our 
company. 
Aviogone 150 mm. 
4500 m. above sea level 
mountaneous from 400 
m. to 1500 m. 
12 models. 
Autograph A7 (with- 
out absolute orienta- 
tion.) 
6) Point transfer device KRP—60 (proper name 
of our device) 
7) Errors at control points 
x, y ( 8 points) m.e. 50cm. max. 123cm. 
z (13 points) m.e. 83cm. max. 264 cm. 
8) Transformation second order 
9) Note: This method is especially powerful 
compared to the usual method of 
adjustment, since the deviations of 
plane coordinates of any point caused 
by its elevation are corrected by 
equation (2) automatically. 
1) Camera 
2) Fleight height 
3) Terrain form 
4) Strip length 
5) Measuring machine 
3. Relative orientation. 
Following to Jerie’s equations” for given 
rotational elements of a photograph x, v and o, 
measured photographic coordinates (x, y) are 
transformed as follows, 
x® =xcosk+ysink, — y? ——xsin «c y cos « 
; x —ftan mp y'" seco 
y Ti y ey mmn (8) 
1+ f -tan o ld f tan o 
"E xe sec o ; (e) — y —f tan i 
X — ye , Jy 5 y 
RE 7 -tan e 14 f tan o 
If we choose x, and @,, Of a left camera and «,, 
q, and o, of a right camera as the five orientation 
elements, their approximate values can be obtai- 
ned by solving normal equation of the parallax 
equations for more than five points. In this case 
parallax equation is, as well known, 
3) Jerie, H. G.: “A contribution to the problem of analytical 
Mar. 1956. 
X 2 Vs „2 
7 phase + T Eg n Yo: 
  
—x Ke 
= Yı—)ı (9) 
If we put into equation (8) the approximate 
values of rotational elements thus obtained, we 
can compute the photographic coordinates of any 
point transformed by the rotations of camera. 
The computations for succesive approximation 
should be repeated until the differences of trans- 
formed photographic ordinates y, and y, of both 
camera for any point will be sufficiently small. 
Relative orientation is thus completed.*^ 
Three dimentional coordinates (x, y, 2) of any 
point can be computed using the photographic 
coordinates of both camera (x,, y,) and (x; ys) at 
the final state of completed relative orientation. 
The equations are as well known, 
X= Xp 
X1— X4 
ay 
y= s. b (10) 
Wa 
X1— X3 
where b is the base length of the model, coordi- 
nate origin is at the center of projection of left 
camera and z is measured downward. 
If we take b as a unit of length, the equation 
(10) becomes 
X1 
Xi— X3 
a 
yum kn (11) 
4. Succesive orientation. 
In our methd, succesive orientation can be de- 
fined as a three dimentional rotation and a scale 
transfer of one of two adjacent models, so as the 
space orientations of their common photograph 
and their scales should coincide with each other. 
In article 3, the relative orientation was executed 
by computation always referring to the coordinate 
system fixed in space at which origin the projec- 
tion center of the left camera had been brought 
on, and the model base taken as a unit of length 
had been the x-axis of the coordinate system itself. 
In the case of relative orientation for the next 
model, coordinate system is the same as before, 
except that the new coordinate origin which is 
the projection center of the left camera of the 
second model should be on the projection center 
  
  
  
aerial triangulation," Photographic Engineering 
4) This method for relative orientatian is not affected by terrain form.