20
Summary.
The results of the partial calibrations of aerial photo
graphs under operational conditions have shown:
1. There have been found considerable differences be
tween the radial distortion of the photographs and of the
lens according to laboratory calibration.
2. It is therefore of obvious importance to calibrate, at
least partially, aerial photographs under operational con
ditions in order to be able to apply corrections in connec
tion with the reconstruction of the bundles of rays. The
corrections may be different for analogue and analytical
work, depending upon the nature of the systematic errors.
3. There exists in practically all tested aerial photographs
a considerable weight variation between different parts of
the photographs. The weights can be expressed at least
approximately as functions of the radii from the principal
point. This is of importance, primarily for analytical resti
tution.
4. Considerable improvement of the basic accuracy has
been found in photographs on polyester base in compari
son with acetate base. The root mean square values of the
standard errors of unit weight have been found in the
relation approximately 5 : 8. This is equivalent to a linear
difference of approximately 7 micrometers. The conse
quences of this fact particularly for the selection of plott
ing instruments and methods are evident.
5. It is possible to correct certain systematic errors of
the image coordinates from the discrepancies in fiducial
marks. It would be of value to use at least 8 fiducial marks,
the accuracy of which must be high. Still more valuable
would be at least one more given point if it can be located
in the neighbourhood of the principal point.
6. One reseau camera has been used for partial calibra
tion described here. No significant improvement of the
accuracy of the final image coordinates has been found.
More similar tests of such cameras of different makes are
desirable.
7. The comparison between the standard errors of unit
weight of image coordinates and the standard errors of
unit weight of corresponding y-parallaxes showed a con
siderable correlation effect, the correlation coefficient
amounting to about 0.60, and the rectangular weight
coefficient to about 0.70. Taking this fact into account it
should be possible to estimate the basic accuracy of image
coordinates from the determination of the standard error
of unit weight of y-parallaxes which can be done in any
arbitrary pair of overlapping photographs. The standard
error of unit weight of image coordinates has been found
to be about 1.4 times the standard error of unit weight of
the j-parallaxes. The errors of the measuring instrument
(stereocomparator Wild StK 824) can be neglected in this
case.
8. It is highly desirable that the geometrical quality of
aerial photographs be checked at suitable intervals and
under operational conditions. Tolerances can be established
according to ordinary principles.
3.12 Results of Partial Calibrations of Terrestrial
Photographs
With the test fields, briefly described above, and mea
surements of image coordinates in the stereocomparator
Wild StK 824 a great number of partial calibrations of
photographs from various cameras such as phototheod
olites, stereocameras, ordinary amateur cameras, X-ray
instruments etc. have been performed. Because of the limited
space here, only a few examples of all results will be shown.
See Figs. 3.12: 1—16. In the summary some general points
of view, gained from the experiences had in this work,
will be extracted. In particular, cameras tested for use
in medical applications of photogrammetry have given
interesting results proving the necessity of careful in
vestigations of such cameras for purposes where the
quality of the results can be of particular importance.