m
ke
re
is
in
or
ta
ni
ng
'SS
re
Th
las
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B6. Istanbul 2004
industry and users like national mapping agencies joined the
network.
2.2 Objectives
The objective of this Digital Camera Calibration project is two-
fold:
— Collection of publicly available material on digital airborne
camera calibration to compile an extensive report
describing the current practice and methods (Phase 1).
— Empirical testing with focus on the development of
commonly accepted procedure(s) for airborne camera
calibration and testing, based on the experiences and
advice of individual experts (Phase 2).
As a result of Phase 1 a report will be compiled based on the
support of all project participants, i.e. camera producers and
users. Such summary will help to create a common knowledge
base for the formulation of future strategies and later
experimental work in Phase 2. This status report is helpful for
digital camera system users to gain their experience with digital
camera calibration aspects. Furthermore, this report should list
open problems which need to be solved. All participants of the
first initiate meeting welcome the idea, that this report is open
to producers, users and customers.
The second phase should focus on the development of
commonly accepted procedure(s) for camera calibration and
testing. It seems to be necessary to concentrate on some of the
technical aspects in a sequential order, starting with geometrical
aspects and verification in a limited number of test flights by
different camera producers and discussion on radiometric and
image quality aspects. One aspect is the design for optimal
calibration flight procedures to be tested then empirically.
Another aspect is a collection of recommendations of producers
on how customers should calibrate and do the processing. It
requires a fine definition of goals which should not lead to
direct comparisons of cameras, but to individual
recommendations for each major camera type. The definition of
goals and the design of empirical tests has to be discussed based
on the report compiled in Phase |.
It has to be mentioned that the project itself will focus on the
calibration of digital airborne camera systems only. The
combination of LIDAR and imaging sensors is not considered
since this is a registration and no calibration problem.
3. ASPECTS OF CAMERA CALIBRATION
3.1 Definitions
Before focussing on the topic “Digital Camera Calibration” by
presenting the applied methods for three digital systems, the
general aspects of traditional camera calibration as mentioned in
the Manual of Photogrammetry (Slama 1980) are briefly cited
in the following:
— Camera calibration is the process, whereby the geometric
aspects of an individual camera are determined.
- [tis performed in the order that the photo obtained with the
camera is used to produce maps, two allow measurements,
whereby ground distances or elevations can be obtained
and to make orthophotos.
— lt is possible to perform calibration to some order on any
camera, but the cameras used to obtain the most accurate
geometric data are specially designed for that purpose
(namely high-quality lenses, usually at infinity focus).
High-quality includes both well defined images and
accurate positioning of the image, large aperture possible
205
without introducing excessive distortions, special
geometric features like fiducials for determining a
‘coordinate system and for controlling the film behaviour.
— Calibration assumes, that the thing being calibrated is
stable between calibrations.
— Calibrated values and their accuracy are reported in a
camera calibration certificate with tables and graphs.
Although most of these definitions are generally valid for all
types of cameras (i.e. analogue and digital), some remarks
should be given related to digital sensors: As already mentioned
the multi-spectral capability is one of the major selling points
for the new digital sensors, hence the calibrations should not
only be restricted to the geometric aspects but to the radiometry
part also. Traditional calibration only focuses on the geometry
task. The photo interpretation application, which obviously is of
increasing future importance, is not considered — especially
when thinking on the small to medium format digital sensors
non dedicated for airborne use but increasingly used to obtain
fast and coloured images for applications in monitoring of land
use changes, disaster and risk assessment, forestry and others
like real estate search and promotion or tourism. Additionally,
those sensors are not specially designed for highest accuracy
evaluation which directly covers the point of stability between
calibrations. Finally, there is no definition or standard on how
the calibrations should be documented.
Since there are different techniques to perform camera
calibration the Manual of Photogrammetry (Slama 1980)
divides between two basic methods. Their difference is due to
the fact, whether the reference values for calibration are
presented in object or image space:
— Present an array of targets at known angles to a camera
which records their images. The targets may be optical
stars (simulating infinite targets) or terrain targets imaged
from towers, aircraft or ground. The recorded images are
measured and the data reduced from the measurements
provide the elements of interior orientation. Many physical
controls are necessary.
— Clamp a master grid in the focal plane, measure the
observed angles in object space using a visual or
goniometer technique. The distortion is computed from the
focal length and the difference between the image and
object angles.
The parameters of interior orientation are closely related to
camera calibration, since a camera is signed as calibrated if the
parameters of interior orientation are mathematically defined,
namely:
— Focal length f,
— coordinates of principle point x, and y, and
— geometric distortion characteristics of the lens system, i.c.
symmetric radial distortions, asymmetric distortions caused
by lens decentering.
No matter of the applied method, the accuracy of camera
calibration depends on the quality of known geometry of targets
being viewed from the camera. This is the reason for the
complex and costly equipment used for laboratory calibration
methods.
3.2 Laboratory calibration
From classical photogrammetric point of view the laboratory
calibration is the standard methodology used for analogue
airborne frame sensors. The results of such lab calibrations are
documented in the well known calibration certificates. In order
to verify the validity of calibration parameters, this calibration is
repeated within certain time intervals, typically each two years.
