2
with the mathematical model in the calibration.
Test Fields
There are several laboratory methods for calibrating photogrammet-
ric cameras. Especially designed instruments are very often
necessary, such as goniometers, collimator banks, multicollimators.
In addition to this there are field methods for camera calibration.
Most laboratory methods and many of the field methods are intended
for cameras that are focussed on infinitely distant objects. The
methods are not directly applicable for cameras focussed at short
distances. The goniometers and collimators have to be differently
focussed and in some field methods the location of the exterior pro
jection center must be known before the calibration can begin. If the
position of the exterior projection center is defined in the system of
the test field by coordinates that are in error, this introduces errors
in the determination of the principal point and the camera constant.
This has been the reason why calibrations of close-up cameras
using, for example, the grid method, only contain a determination
of the radial distortion and the standard error of unit weight of the
image coordinates. As the origin of the trouble is the unknown
position of the exterior projection center this difficulty would be
eliminated if the position was determined in the calibration itself.
The grid method has a plane test field, which as a rule is parallel to
the image plane. The determination of both the interior and the
exterior orientation is not possible in this case. The corresponding
normal equations have a non-singular solution. To overcome this, it
is necessary to extend the test field into three dimensions.
Three-dimensional Test Fields
From the viewpoint of theory of errors, it has been regarded as an
advantage to calibrate cameras under conditions that are as similar
as possible to those of ordinary work. Firstly there can be regular
differences in the elements of interior orientation between laboratory
and field calibration. Secondly the accuracy will be determined under
operational conditions. Short range photogrammetric cameras are
used to determine object coordinates in a three-dimensional space.
It seems to be a consequence that the test field should be of the
corresponding shape and size. The standard errors of the principal
point and the camera constant decrease with decreasing standard
error of unit weight, decrease with increasing opening angle of the
bundle of rays, and decrease with increasing extension of the test
field in the direction of the camera axis. For a proper estimation of
the accuracy the points should be evenly distributed over the image.
Test objects have been designed with these ideas in mind and used for
calibrations of stereocameras, photodolites, ordinary cameras,
X-ray equipment, cinetheodolites, stereomicroscopes, radio-iso-
topic imaging systems etc. Aerial cameras can be conveniently
checked over mountainous terrain. 5)