Fig. 
6 
Frequency curve of errors at the corners. 
Xo 
&x, AY 6 control poinito 
sol- À £15.59" Y 
Max TOM 
a 
o b 
= —0.000484.* 0,0005 TAX tl.o0 
-b 
a 
C 
d 
-d 
C 
X7 *3331863 t1,0000453,-0.000574 4-000 l4 T (X^-59?) -0,00046x 2x4 
00452.t0.00014 6 (32-4?) - 000014 TX 2XXd 
M = 1.000095 
AM= 0.000206/( x* 4?) 
4o}- RU 
AX aY>~ d 
3 control point vus X= + I.0001T5 tl.000180X -0.0006 741 - 0.000142 (X?- 42?) -0,000153X2XY 
418.00 \ 
cm x 
SE Max 80 N Y= -0.000658 40.000 b79X+1,0001804 10.000153 CC- 3?)-000014 2X2 x 4 
  
21000180 
AM 20.000208 42) 
  
  
20 
total ; 60 
m. 30k 
0 ] 
30 
exvor (c ) 
—> 
Equations show coefficients of conformal transformations for the 2 cases. 
4) 
In the case of above experiment, we found 
that the adjustment of Autograph A7 had not been 
good enough, i.e. x and y inclination had not yet 
been removed sufficiently. 
So that, after carefull adjustment of A7 we made 
the same measurements of grid plate for three 
kinds of space orientation, on both camera of 
A7, respectively. And three models for each 
camera were constructed and connected by the 
combination of these grid plate photographs. In 
this case the discrepancies at common points of 
adjacent models are sufficiently good as follows, 
x y z 
upper point 1 19(+13) 29( 4-19) 
middle point 1 6(— 6) 17(— 8) 
lower point 1 11(+ 8) 20(— 9) 
Unit is 0.00001 of base length, and the 
values shown in bracket are the means made 
under consideration of signs. 
5) Ideal grid plate extension. 
In above cases a grid plate is actually measured 
on Autograph A7, giving some values for rotation 
and inclinations. But such photographs of grid 
plate can be easily made by computations, and 1f 
we make analytical extension of these grid plate 
photographs ideally made by computation, we ob- 
tain an theoretical strip. Example of three model 
extension will be shown in table 1. In table 1, 
(x, y, 2) are the strip coordinate before absolute 
orientation. So that the strip remains inclined, 
unit of length is the base distance of the last 
model and origin is at the projection center of 
the right camera of the last model. 
A computing measurements of photographic co- 
ordinates are made as follows. First, we make the 
photographic coordinates down to the order of 
0.001 mm. by direct computation. And next, we 
make double reading values which have two deci- 
mal figures below mm., so as the mean value of 
these double readings be nearest to the true value. 
Then it is clear that errors of the photographic 
coordinates thus obtained are not greater than 
0.0025 mm. 
Table 1. 
NAME X Y Z 
0001  —03.000073  — 00.942481 +01.904517 
0002  —03.000070  --00.057066  --01.874098 
First 0003  —03.000067  4-01.056660  --01.843655 
model. 0004 —02.000050  — 00.942505 +01. 904737 
0005 —02. 000035 +00.057067 +01. 874240 
0006 —02. 000069 +01. 056613 +01. 843751 
0004  —02.000066 —00. 942511 +01. 904608 
0005 —02.000049 +00. 057051 F01. 874151 
Second 0006 —02. 000072 +01.056677 +01. 843853 
model. 0007 —01.000054 —00.942480 +01.904564 
0008  —01.000049  4-00.057073  -4-01.874175 
0009 —01.000021  --01.056648  --01. 843856 
0007  —01.000038 —00.942505 +01. 904616 
0008 —01.000049 +00. 057051 +01. 874226 
Third 0009 —01.000000 +00. 056603 +01. 843747 
model. 0010 00. 000000 —00.942542 +01. 904664 
0011 00. 000000 +00. 057040 +01. 874273 
0012 00. 000000 . 4-01. 056592 . -- 01. 843793 
  
      
   
   
   
    
  
   
   
  
   
   
   
    
  
   
    
    
     
      
     
   
   
   
   
   
       
    
        
   
       
     
  
        
         
  
co -3 c C ui 
te 
CÁC 
rer 
the 
ph: