Principal Point [pixels] 
Frame 1 4 9 91 
PP PP.dev .PP .dee PP dev 
  
  
Posix 77  -85 -03 67 +10 88 11 
Posly 23" 442465 99-22 +11.1 +133 
Pos2x -9.5 -99 -0.4 -9.5 00 .-100. -05 
Pos2y -1.6 0.0 +1.6 +7.2 +88 | 49.1 «410.7 
   
    
   
Scale factor between x and y [x/y] 
  
  
  
table 3 Principal Point 
5.3.3 Radial distortion coefficients The radial 
distortion coefficients are stable and does not seem to 
be effected by the moving principal point. In this test 
they are calculated separately in x and y. In this 
camera type, with the film fixed during the exposure 
and not a rotating prism, it did not improve the 
result compared to a common parameter in x and y. 
The deviations referres to the first frame. 
  
Frame 1 4 90 91 
SF SF SF SF 
dev dev dev 
Pos 1 0.9986 0.9973 0.9998 0.9999 
-0.0010 +0.0012 +0.0013 
Pos 2 0.9993 0.9997 0.9993 1.0002 
+0.0004 0.0 +0.0009 
  
  
  
Radial Distortion separate in x and y [polynomial] 
  
Frame 1 4 90 91 
RD RD RD RD 
dev dev dev 
Posix -.35e-7 0.37e-7 -0.34e-7 -0.35e-7 
-0.2e-9 +0.1e-9 0.0 
Posly -33e-7 -0.31e-7 -0.34e-7 -0.35e-7 
+0.2e-9 +0.1e-9 0.0 
Pos2x -32e-7 -0.31e-7 -0.32e-7 -0.31e-7 
+0.1e-9 0.0 +0.1e-9 
Pos2y . -32e-7 -0.31e-7 -0.30e-7  -0.32e-7 
+0.1e-9 +0.2e-9 0.0 
  
  
  
table 4 Radial Distortion 
5.34 Scale difference in x and y The scale 
difference between the x and y axis is very small, less 
than 0.3 percent, but might be more significant for 
other types of cameras. 
table 5 Scale Factor 
5.3.5 Position and rotations The position and 
rotations of the camera are no actual calibration 
parameters, but is calculated by the program and used 
as approximate values for the bundle adjustment if 
the test run is performed with the same set-up as the 
calibration. In this study they are compared to see if 
the camera moves during the exposure of the film. 
The position has only very small deviations, less 
than 5 mm. The rotations have deviations which 
may be of more importance, less than 0.1 grade, 
which may be caused by the rather heavy forces on 
the camera house during the exposure. This is also 
supporting the correctness of recalculating the outer 
orientation and principal point rather often or for 
every frame. 
54 One calibration for whole sequence 
The cameras were calibrated on frame one and an 
outer orientation with no free calibration parameters 
were performed. The orientation parameters were 
then used to calculate 3D coordinates with fixed 
orientation on all four frames. The purpose of the 
test was to find out if it was enough to do one outer 
orientation with the bindle adjustment in the 
beginning of the strip and then use these parameters 
for the rest of the strip. 
The comparative measure is the mean radial 
deviation (MD) from the known points in the test 
field. The unit for the deviations is in [mm] and the 
number of points used is 60. The result shows that it 
is not sufficient to do one outer orientation in the 
beginning of the sequence and use it for all following 
frames. 
  
Deviations, one outer orientation [mm] 
Frame 1 4 90 91 
MD MD dev MD dev MD dev 
  
Pos 1+2 6.8 78 410 2741 +20.3 25.8 419.0 
  
  
  
table 6 One calibration for whole sequence