4-5-1 
SENSOR INTEGRATION AND CALIBRATION OF DIGITAL AIRBORNE THREE-LINE CAMERA SYSTEMS 
Michael Cramer, Dirk Stallmann and Norbert Haala 
Institute for Photogrammetry (ifp), University of Stuttgart 
Geschwister-Scholl-Straße 24D, 70174 Stuttgart / Germany 
e-mail: Michael.Cramer@ifp.uni-stuttgart.de 
Commission II, Working Group 1 
KEY WORDS: GPS, INS, aerial triangulation, push-broom line scanners 
ABSTRACT 
The determination of the exterior orientation parameters is an essential pre-requisite for the evaluation of any imagery from terrestrial, 
airborne or satellite based sensors. Normally, this georeferencing processing is solved indirectly by using a number of well known ground 
control points and their corresponding image coordinates. Using a mathematical model for the relation between image and object space 
the exterior orientations can be calculated and the local image coordinates are related to the global ground coordinate system. In principle 
this approach can be applied for georeferencing of push-broom line scanner imagery, but this process is highly inefficient. Due to the 
large number of unknowns a large number of tie and control points is necessary for orientation determination. To allow an operational 
processing the direct measurement of exterior orientation using GPS and INS and additional information is inevitable. Within this article 
the geometric processing of high resolution line scanner imagery is described and the test results from different airborne test flights flown 
in 1998 are given. 
KURZFASSUNG 
Die Bestimmung der Parameter der äußeren Orientierung ist eine wichtige Voraussetzung für die Auswertung terrestrischer, luft- oder 
weltraumgestützter Bilddaten. Normalenweise wird diese Georeferenzierung indirekt durch die Verwendung bekannter Paßpunktinforma 
tionen am Boden und die Messung der zugehörigen Bildkoordinaten gelöst. Unter Verwendung eines mathematischen Modells für die 
die Beziehung zwischen Bild- und Objektraum können die äußeren Orientierungen berechnet und die lokalen Bildkoordinaten in Bezug 
zu dem globalen Geländekoordinatensystem gebracht werden. Prinzipiell ist dieser Ansatz der Georefernzierung auch auf Pushbroom- 
Zeilenscanner-Daten übertragbar, allerdings ist dieser Prozeß hochgradig ineffizient. Wegen der hohen Anzahl von Unbekannten wird 
für die Orientierungsbestimmung eine große Zahl von Verknüpfungs- und Paßpunkten benötigt. Im Hinblick auf eine operationeile Ve 
rarbeitung der Daten ist daher die direkte Messung der äußeren Orientierung mittels GPS, INS und weiteren Sensoren unvermeid 
bar. In diesem Artikel werden die geometrische Auswertung hochaufgelöster Zeilenscanner-Daten beschrieben und die Ergebnisse 
verschiedener 1998 durchgeführter Testflüge vorgestellt. 
1 INTRODUCTION 
Up to now the analogue acquisition of image data prevents pho 
togrammetry to become a fully digital, towards real time mapping 
system. Todays systems for digital airborne image acquisition can 
be split into frame and push-broom systems. Despite the ongo 
ing progress in the development of airborne frame cameras it still 
seems to take some more years to replace the large format film 
based cameras with equivalently sized digital frame systems. The 
maximum resolution of digital frame sensors available is about 
9000 x 9000 pixel. Assuming 10^m pixel size, this sensor cov 
ers about 80cm 2 , which is still significantly less compared to the 
standard photogrammetric analogue image format of 23 x 23cm 2 . 
Today, digital systems using the line scanning geometry are the 
only imaging sensors that can compete with digitized aerial photos 
in terms of acquired area and image resolution. 
These line scanners can be expanded to multi-line sensors provid 
ing stereoscopic and multi-spectral data simultaneously. These are 
enormous advantages compared to traditional analogue data. Un 
fortunately line scanning systems are affected by one major fact: 
Georeferencing of image data is more complex compared to stan 
dard aerial triangulation. Although the traditional indirect approach 
using ground control points for the determination of the exterior 
orientation of the images works for airborne sensors, this process 
is highly inefficient. For line scanner systems a direct processing 
strategy utilizing direct measurements of the exterior orientation 
provided by satellite (GPS) and inertial navigation system (INS) 
is necessary for operational and efficient data evaluation. Even 
though direct georeferencing is no must for digital frame cameras 
a GPS/INS component is also included in some systems (Toth, 
1998). 
Within this article the integration of GPS, INS and line scanning 
imagery for the georeferencing of a digital airborne line camera 
system is shown. Following a short discussion of different ap 
proaches of georeferencing of image data the combined approach 
using GPS, INS and measurements from image space in an ex 
tended aerial triangulation process is described (section 3). Com 
pared to the stand-alone GPS/INS integration the combination with 
image observations increases the reliability of the whole sensor 
system. Remaining systematic effects can be modeled using addi 
tional parameters similar to self-calibration. Furthermore, the pho 
togrammetric constraints are used to eliminate the systematic INS 
error effects significantly. The influence of the different error types 
which are introduced with the different sensors are shown. Special 
focus is given on the effects caused by systematic INS errors. The 
influence of these errors is shown in some simulations (section 4). 
In the last part the functionality of the combined aerial triangulation 
algorithmn is presented. The practical results of different testflights 
using different camera systems over a well known testfield close to 
Stuttgart/Germany with more than 150 signalized check points on 
the ground are given. 
2 PRINCIPLES OF GEOREFERENCING OF IMAGERY 
The determination of the exterior orientations is a major task in 
the evaluation procedure of image data and can be done using 
different orientation methods. These methods can be classified 
in indirect- or direct approaches and are applicable for traditional 
frame (digital/analogue) or line imagery. Table 1 gives a short