METRIC OR NON-METRIC CAMERAS 
image quality, which is intolerable for pre- 
cision measurements. Nevertheless the dia- 
gram can be used for other values; the result- 
ing depth of focus has only to be multiplied 
by the factor to the initial values of the 
aperture stop or the diameter of the circle of 
confusion. 
The accuracy estimation has been per- 
formed with a fictitious test field of eight 
control points. These points were located in 
the corners of a quadratic prism (see Figure 
2). The side length of the prism should be 
chosen so that it might fill four fifths of the 
picture format. The height of the prism is 
variable and is expressed as a fraction of the 
imaging distance. The computation was 
stopped when the depth of the fictitious test 
field was equal to its lateral extension. 
Finally a figure for the mean square error 
of unit weight e, must be assumed in order to 
compute the variances for the principal point 
and the principal distance. This term o, 
indicates the measuring precision in the 
picture and can be determined from the 
residual errors of a camera calibration. Ac- 
cording to various experiments*5.8 a measur- 
ing accuracy of 0, = = 3 um can be achieved 
with glass plates or plane film, and for roll 
film ao, = = 10 to 12 um must be expected. 
The simulated camera calibration was per- 
formed on an electronic computer (CDC 
6400). The program was run for various focal 
lengths and opening angles of the camera. 
The computation showed that the accuracy 
requirements for the principal distance and 
[m] | D 
   
44 AZ/Z 
1 
1%0 7 
Fic. l. Ratio between the depth-of-focus and 
the maximum object distance AZ/Z. This value 
can be used to enter in the graphs of Figure 3 and 
Figure 4. (admitted circle of confusion, 30 pm; 
aperture 1:16; D, focusing distance) 
107 
B 
FTN 
\ 
\ 
  
AZ 
  
  
  
  
> | 
  
  
ho at 
3 L 
em. NS 
SS 
> 
— 
  
ue. | 
L 
  
  
  
| 
| 
| 
| 
| 
/ | \ 
| 
| 
| 
| 
| 
Fic. 2. For the determination of 
tolerances for the principal dis- 
tance and the principal point, a 
simulated camera calibration is 
used. The figure shows the distri- 
bution of the control points and 
the dimensions of the fictitious 
test field. Normally the test field 
has the shape of a cube. For close- 
range photographs the depth-of- 
focus can get very narrow. Then 
the depth extension AZ of the test 
field has to be adapted to the 
depth-of-focus (a is the angle un- 
der which the test field is photo- 
graphed, Z the maximum distance 
of the object points). 
the principal point are independent of the 
focal length and depend only on the opening 
angle of the camera and the extension of the 
testfield. This means that the numerical 
values for the tolerances of the inner orienta- 
tion would be the same for a small-size 
camera as for an aerial camera with a picture 
format of 23 x 23 cm as long as the opening 
angle of the lenses are the same. 
The computed variances are presented in 
Figures 3 and 4. The graphs show that a high 
precision for the parameters of the inner 
orientation is necessary only for objects with 
considerable depth extension. If the object is 
fairly flat, then the tolerances for these pa- 
rameters are obviously less severe. In the 
extreme case, for a completely flat object and 
vertical photographs, the values for the prin- 
cipal distance and the principal point can be 
arbitrary. Deviations from the nominal val- 
ues are completely compensated by a proper