110 
same estimations should be valid for the 
larger format cameras such as the Linhof- 
Technika and the Phototheodolite. The vari- 
ation of the principal point also should be 
approximately twice the value for the prinei- 
pal distance. The reproducibility of the prin- 
cipal distance seems slightly less accurate 
according to the figures in Table 1, which 
might be due to deformations of the image 
plate. 
A comparison of these figures with the 
tolerances discussed earlier shows that the 
reproducibility of the principal distance is 
rather satisfying whereas the precision of the 
principal point is adequate only for an as- 
sumed measuring accuracy of a, = + 10 um. 
This is valid for non-metric cameras and to 
some extent also for the tested metric cam- 
eras. Again, the tendency can be seen that 
narrow-angle cameras are better fitted for 
precision measurements than wide-angle 
cameras. Although the reproducibility of the 
principal point becomes less accurate for 
cameras with long focal lengths, the toler- 
ance for this parameter increases even more 
rapidly. 
MODEL ACCURACY 
The accuracy discussion has assumed that 
the measuring precision should be limited 
only by uncontrollable components, such as 
film shrinkage or lack of flatness of the 
photographic plate, whereas errors of the 
orientation elements should be small enough 
to be neglected. Up to now the discussion has 
concentrated on the elements of inner orien- 
tation, but would be incomplete without the 
inclusion of elements of exterior orientation. 
In the following, an example is given in 
which the overall precision was decisively 
limited by vibrations of the stereocamera 
'TABLE 1. 
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1976 
whereas the use of metric cameras would not 
have contributed to any increase of the 
measuring precision. 
At the Swiss Forest Research Institute an 
investigation has been undertaken to per- 
form a forest inventory with extremely large- 
scale photographs taken from helicopters. 
The task was to measure tree height, stem 
diameter, and a few other parameters in 
deciduous woods with the help of stereo- 
photographs. The required precision was = 1 m 
for the tree height and + 1 to + 2 cm for the 
stem diameter. The base length chosen could 
be relatively small because the precision 
requirements in depth were not very severe 
and the use of a stereocamera became fea- 
sible. 
After a few experiments, a base of 4.5 m 
was adopted, and two Hasselblad 500 EL 
cameras with Planar 1:2.8/80 mm lenses 
equipped with film were used; the flying 
height chosen was 100m (see Figure 6). It 
was of great importance that the model scale 
be kept constant within about + 1 per cent 
because a signalization of control points 
seemed unreasonable. The observation of 
this tolerance proved to be very difficult. 
Theoretically, scale errors can be caused by a 
varying base length, an inferior definition of 
the principal point (for the indicated meas- 
urements, the principal point used for the 
reconstruction of the pencil of rays must not 
exactly coincide with the principal point of 
autocollimation), and by a variation of the 
angle of convergency of the two cameras 
during flight (see Figure 7). To avoid scale 
errors in the model of more then 1 per cent, 
the tolerance for the base length would be 
+ 5 cm, or = 30 to 40 um for the definition of 
the assumed principal point (picturescale 
1:1200) and + % for the angle of convergency. 
The observation of the convergency proved 
COMPARISON OF THE TOLERANCES FOR THE PRINCIPAL POINT AND THE PRINCIPAL DISTANCE 
WITH THE REPRODUCIBILITY OF THESE PARAMETERS FOR PLATE CAMERAS*S, (A: NON-METRIC CAMERAS, 
M: METRIC CAMERA, @ OPENING ANGLE COMPUTED FOR A REDucED PLATE FORMAT OF 80%, THE 
TOLERANCES IN BRACKETS ARE VALID FOR A DEPTH OF FiELD or Az/z = 1/10). 
  
Principal Point 
Principal Distance 
  
  
  
  
Camera a Tolerances Reprod. Tolerances Reprod. 
g,7 -lÜum 0,= +3 yum g,=*10 um 0,= +3 um 
Hasselblad 500 C (A) 34* 65 20 70 85 25 31 
f = 80 mm, 5.5 x 5.5 cm (100) (30) (290) (87) 
Linhof Technika (A) 36° 60 18 70 80 24 50 
f = 135 mm, 9 x 12 cm (95) (28) (270) (81) 
Photo-Theo (M) 37* 55 17 46 75 22 32 
f = 190 mm, 13 x 18 cm (90) (27) (260) (78) 
SMK (M) 695 15 10 27 8 10 
f = 60 mm, 9 x 12 cm (75) (22) (115) (34)