6
2. Introduction
j
Most procedures in photogrammetry are based upon the assumption that the
images are central projections of the objects. In many procedures the bundles
of rays of the central projections are reconstructed. The central projection is
also the basis for the projective relation between object and image. Calibration
of pictures, cameras and projectors is a determination of some properties of the
corresponding central projection. In this thesis the calibrations concern only
the geometric properties of the images. Image quality described in terms of re
solution or contrast transfer is not treated. The geometric properties of the cent
ral projection, that are necessary to make it possible to reconstruct the bundle
of rays, are given by the elements of interior orientation. Calibration of came
ras, projectors and pictures is thus of basic importance for almost every photo-
grammetric procedure.
The methods to determine interior orientation can be divided into two groups,
namely, a laboratory group and a field group. In the laboratory the determina
tion can be done under well controlled and ideal conditions, while the field
methods can give the data under operational conditions. The International So
ciety for Photogrammetry, ISP [5], recommends laboratory procedures for cali
brating cameras. These recommendations are suited to aerial cameras and colli
mator techniques. Thus they apply to cameras focussed on infinity.
Using field methods for calibrating cameras provides some advantages.
1. They can be used for close-up cameras focussed on finite distances.
2. They provide the interior orientation and the accuracy of the image co
ordinates under operational conditions.
3. They can check a laboratory calibration under operational conditions.
4. They usually do not require so expensive instrumentation as laboratory
methods, but instead, demand test fields.
The calibrating methods described in this thesis aim at a determination of
the elements of interior orientation, their accuracy and the accuracy of the ima
ge co-ordinates in one and the same procedure. This is done by photographing
three-dimensional test objects, measuring the image co-ordinates and compu
ting the desired quantities on the basis of the method of least squares. In order to
get a representative sample of the image co-ordinates and to increase the num
ber of degrees of freedom, the points should be uniformly distributed over the
image area and so many that it is possible to test hypotheses on the elements
of interior orientation and on the distribution of the residual errors.