KINEMATIC MULTI-SENSOR SYSTEMS FOR CLOSE RANGE DIGITAL IMAGING
K.P. Schwarz* and N. El-Sheimy**
* Professor, Department of Geomatics Engineering, The University of Calgary, Canada
** Senior Engineer, R&D Department, GEOFIT Inc., Canada
ISPRS Commission V, WG III
KEY WORDS: Multi-Sensor Systems, Geomatics, The Global Positioning System (GPS), Inertial Navigation Systems (INS),
Digital Images. Georeferencing, Synchronization. Real-Time Quality Control.
ABSTRACT :
Major progress has been made in close-range digital imaging over the last few years in terms of sensor resolution, data rate, and
operational flexibility. Thus, the usc of such sensors in kinematic applications has become very attractive. To move from the static use
of such sensors to kinematic applications, exterior orientation parameters for the sensor are needed in high dynamics situations. In
exchange. fully automated data acquisition at high speed can be obtained and the range of applications can be considerably extended.
Two basic solutions to the exterior orientation problem are currently available. In the first one, the parameters are determined directly
by using suitable position and orientation sensors. In the second one, they are determined indirectly by extracting them from a block of
images with a sufficient number of known control points. In the first case the svstem is more complex, in the second, the operational
restrictions arc more scvere. In this paper thc emphasis will be on the first approach, the direct determination of sensor position and
orientation, which requires an integrated multi-sensor system.
The presentation will cover both, the concept of multi-sensor integration and implementation aspects. Based on experience with a
number of different systems. features common to most systems will be identified and a unified model for multi-sensor integration for
close range digital imaging will be formulated. Suitable observables for this model will be assessed, and factors affecting system
performance will be discussed. All major features will be illustrated by examples. Finally, data flow optimization and the potential for
automation of the data acquisition and fcature extraction process will be reviewed with a view to future systems.
1. MULTI-SENSORS SYSTEMS AS A TOOL IN
GEOMATICS
Multi-sensor svstems have become an emerging trend in
geomatics because they allow a task-oriented implementation of
geodetic concepts at the measurement level. Examples of such
svstems can be found in airborne remote sensing, airborne
gravimctry. airborne lascr scanning. and mobile mapping from
vans and trains. All of them have in common that the sensors
necessary to solve a specific problem are mounted on a common
platform. Bv synchronizing the data streams accuratelv, the
solution of a specific problem is possible bv using data from one
integrated measurement process only. The post-mission
integration of results from a number of disjointed measurement
processes and the unavoidable errors inherent in this process are
avoided. This results in greater conceptual clarity, task-oriented
system design and data flow optimisation, and also offers in most
cases the potential for real-time solutions which are becoming
more important in many applications.
The trend towards multi-sensor systems in geomatics is fuelled
by the demand for fast and cost-effective data acquisition and by
technological developments which allow to satisfy this demand.
Two developments are especially important in this context:
Digital imaging and precise navigation. Digital imaging sensors
considerably reduce the data processing cffort by eliminating the
digitizing step. They also open the way towards new and flexible
designs of the processing chain, making ample use of
mathematical software tools readily available. In the form of
digital frame cameras, they are
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
inexpensive enough to make redundancy a major design tool. In
the form of pushbroom scanners, they provide additional layers
of information, not available from optical cameras.
Precise navigation has developed to a point where it can provide
the solution of the exterior orientation problem without the use of
ground control points (GPC) or block adjustment procedures.
Since results arc available in digital form, data fusion with the
imaging data is easy and real-time applications are possible in
principle. Operational flexibility is greatly enhanced in all cases
where a block structure is not needed for other reasons. Costs are
considerably reduced, especially in areas where little or no
ground control is available. Current accuracy is sufficient for
many mapping applications, sec for instance Schwarz (1995).
The potential to solve even high-accuracy cadastral applications
is certainly there.
Combining these two developments the concept of the
georeferenced image as the basic photogrammetric unit emerges.
This means that each image is stamped with its georeferencing
parameters, three positions and three orientations, and can be
combined with any other georeferenced image of the same scene
by using geometry constraints, such as epipolar geometry or
object space matching. This is a qualitatively new step because
the georeferencing parameters for each image are obtained in a
direct wav by independent measurement. This is conceptually
different from the notion that a block of connected images and
sufficient ground control is needed to solve the georeferencing
problem. This indirect solution of the problem is currently the
standard procedure and is even used in cases where
georeferencing information from GPS is employed. It is viewed
as €
sup]
geor
cont
gcot
It is
uma;
pus!
In tl
mob
illus
imp
disc
Syst
min
The
cha]
VIS
obje
(19%
Vi
a
Ww
Ct
Ct
St
2. (
The
clos
sele
syst
flex
moc
vehi
acq
sens
prot
gen
acci
navi
Sys
mot
com
use