lack, to an increasing extent, point density with increasing distance
from the format center. Stellar calibrations provide a high point
density and good uniformity in point distribution but fail to provide
information in the format corners (Point distributions F and G).
This is not caused by a lack of suitable object points but rather by
the reduction in resolving power of the lenses.
The different point distributions affect different weighting in
the determination of the parameters of the interior orientation
because of their influence on the optimization of the average radial
lens distortion. It is for this reason that it is planned to compare
all results after they have been used to determine the lens distortion (0
values for each point of the camera reseau, that is, for the point
distributions H shown in Figure 8. The interpolation of the values
for the reseau points will be carried out in stages: firstly, the
radial symmetrical component of the lens distortion will be
determined; secondly, the decentered lens distortion will be
determined from the residual lens distortion remaining after the
subtraction of :he radial symmetrical distortion; thirdly, the
residual lens distortion remaining after the subtraction of both,
the radial symmetrical and the decentering components, will be
transferred from the original data points to the reseau points by least
square interpolation; and finally the decentered lens and the radial
symmetrical distortion components will be added at each reseau point.
Once the lens distortion is known for all reseau points, the
calibrated focal length will be modified by applying the same
weighting function to all the different lev: distortion determinations
in the form of a spatial resection.
At present, the calibrated focal lengths vary somewhat. These
variations are to some extent the result of either accuracy limitations «9
or small actual changes as indicated by the results of calibrations 1b,
4b and 8b, which were all carried out at the National Research Council
using the same procedure. Processing of the goniometer data
provided by the two manufacturers (calibration 2a) and Hakkarainen
(calibration 3a) in the same way as for calibrations lb, 4b and 8b
resulted in very similar calibrated focal lengths (calibrations 2b
and 3b). Calibrations la, 4a, 5, 7 and 8a were carried out with the
image plane rather than the reseau plane as reference; all give
similar calibrated focal lengths. The cause of the deviation of
the calibrated focal length for calibration 6 from the mean of all
other calibrations has not yet been determined. This deviation is
particularly pronounced for the RC8 camera; a glance at the
—16-