- 3 -
Example of a Testfield
For stereocameras and similar equipment a test field of steel has
been made at the Division of Photogrammetry, Royal Institute of
Technology, Stockholm. This testfield consists of 179 targets
mounted on 8mm tubes, which extend between two horizontal frames
approximately 2.40 meter apart. This steel cage is so-constructed
and supported that it is free to move as a unit, but without internal
deformation. The targets have a circular pattern with a small central
cross. The size and shape of the circles fit the measuring mark of
the comparator (Wild Stk 824) that is used for measuring the test
photographs. The central crosses were used as targets in the
"geodetic" measurements for determining the X-Y-Z co-ordinates of
the points in the test field. The size of the test field is
3.6x2.4x2.8 metres. The "geodetic" procedure was as follows.
Three theodolite stations were used. Abase net of these stations and
three reference points in the steel cage was measured with the
theodolites used as collimators pointing parallel to each other in order
to avoid eccentricities. Because of the great number of points and
therefore the time involved, because of tripod movements, and
because of the possible movements of the test field, the theodolite
Wild T3 observations were made in sets of 6 to 1 0 points. The time
required for every set was about half an hour. All recordings were
punched directly on cards. In every set three reference base-net
points in the cage were measured together with the 6 to 1 0 other
points. To eliminate systematic errors of the theodolite, erect and
inverted readings were taken immediately in succession without
changing the focus. Every set of measurements was referred to the
theodolite stations of the base net. The actual position of the
instrument in a set was determined by a resection from the three
reference points. The other readings were then referred to the
station of the base net. The absolute scale of the coordinates was
indirectly obtained from the end points of a two meter invar bar,
which were measured with the same procedure. All checking, center
ing, averaging and intersecting of new points were done by means of
a CD 3600 computer, programmed by Ole Jacobi, Copenhagen, Den
mark. The accuracy of the coordinates has been estimated to be
0. 07 mm expressed as a standard error vector in space. During a
two year period the accuracy remained the same, as far as the
standard errors of unit weight of the test image coordinates indicate.
Example of Statistical Analyses of Interior Orientation
A metric camera is defined as a camera that has a known and con
stant interior orientation. The calibrations that are made with the
method described in this paper yield the interior orientation of that
particular picture that is used in the calibration. This is also the
case for almost every calibrating procedure. Before we can talk of
the interior orientation of the camera, we must make sure that the
calibration picture is representative of all the pictures that are