ADVANCES IN CHARACTERISATION AND CALIBRATION OF
DIGITAL IMAGING SYSTEMS
Horst A. Beyer
Institute of Geodesy and Photogrammetry
ETH-Hönggerberg, CH-8093 Zurich, Switzerland
phone: +41 1 377 3042; fax: +41 1 372 0438; e-mail: horst@p.igp.ethz.ch
Commission V
ABSTRACT
Real-Time Photogrammetry is making inroads into various applications due to its ability for on-line and real-time re-
sponse. This opens applications in industrial measurement, robotics, traffic, medicine, etc. The accuracy of these Real-
Time Photogrammetric Systems is a primary requirement for a large number of tasks. It is thus of utmost importance to
investigate the accuracy potential of digital imaging systems. The radiometric and geometric properties of all elements
involved in image acquisition and information extraction must be investigated in order to locate sources of degradations.
With respect to photogrammetric applications effects leading to a degradation of the accuracy of three-dimensional mea-
surements are of primary interest, whereas purely radiometric and geometric deformations which have no effect and/or
which can be modelled are not of concern. It is thus necessary to locate the origins of degrading factors and to quantify
their effect on the accuracy of three-dimensional measurements. This paper deals in a first part with the analysis of the
radiometric and geometric characteristics of a number of components. Their effects on the accuracy of photogrammetric
are discussed. A three-dimensional accuracy test with a testfield is used to demonstrate the accuracy potential of off-the-
shelf hardware. Some limitations found in the test are pointed out and potential improvements are discussed.
KEY WORDS: Calibration, Image Quality, Accuracy, Sensor, Digital Systems
1 INTRODUCTION
Digit imaging system have become an important topic in
Digital Close-Range Photogrammetry. The readily avail-
able data makes real-time and on-line processing possi-
ble. This capability has opened a wealth of new
application areas. The accuracy of these systems plays
amongst many requirements such as automation, robust-
ness, flexibility, and user-friendliness a fundamental role.
Real-Time Photogrammetry was already feasible with
imaging tubes, but their low degree of stability limited
their accuracy potential such that they did not gain wide-
spread use in photogrammetric applications (e.g. Wong,
1969; Pinkney, 1978; Burner et al. 1985). The invention
of the charge-coupled device in 1970 and the subsequent
rapid development of solid-state imaging sensors, the
commercial availability of flash analog-digital-convert-
ers, and the drastic performance increases of processing
systems led to a dramatic change. The new sensors ex-
hibited excellent characteristics and the processing sys-
tems allowed real-time feedback and/or the use of ever
more demanding algorithms. Several characteristics, the
accuracy, and the suitability of solid-state image sensors
for Real-Time Photogrammetry were investigated among
others by Gülch, 1984; Haggren, 1984; Curry et al.,
1986; El-Hakim, 1986; Grün and Beyer, 1986; Wong and
Ho, 1986; Dähler, 1987; Heikkilä, 1988; Lenz, 1988;
Luhmann, 1988; Bósemann et al., 1990; Burner et al.,
1990; Gustafson, 1991; and Shortis et al., 1991. Further-
more a wealth of knowledge on solid-state sensors, their
characteristics, and performance was also acquired in as-
tronomical and space tracking applications (see Beyer,
1992b for an extensive list of references). Pitfalls and the
performance of frame grabbers were also investigated
(e.g. Raynor and Seitz, 1990).
Building on this knowledge and expertise the radiometric
and geometric characteristics of elements involved in im-
age acquisition with solid-state imaging devices and their
effect on the three-dimensional accuracy of photogram-
metric measurements was investigated (Beyer, 1992a).
Figure 1 shows the major elements of imaging and infor-
mation extraction. The discussion shall be based on the
assumption that the three-dimensional position of spa-
tially limited features, such as signalized targets, are to
be determined. This simplifies some evaluations but does
not inhibit any generalization of the conclusions. The
aim is thus to locate sources leading to geometric dis-
placements of feature positions and/or to radiometric
degradations decreasing the measurement precision and/
or inducing geometric displacements.
First the characteristics of the elements are analyzed fol-
lowing the path of the data as shown in Figure 1. The in-
fluence of degrading factors on the accuracy of three-
dimensional photogrammetric measurements is dis-
cussed. Synchronization, target size, temperature, and
other effects are addressed and the inner precision of
geometric measurements is investigated. Finally the per-
formance of an off-the-shelf CCD-camera is investigated
with a three-dimensional testfield. The effects of several
degrading factors on the accuracy are analyzed. The per-
formance of pixelsynchronous frame grabbing and PLL
line-synchronization is compared. Limiting factors and
potential improvements are discussed.
