Christian Heipke
INTEGRATED SENSOR ORIENTATION - AN OEEPE TEST
C. Heipke!, K. Jacobsen!, H. Wegmann!, O. Andersen”, B. Nilsen?
i University of Hanover, Institute of Photogrammetry and Engineering Surveys, wegmann@jipi.uni-hannover.de
“ Agricultural University of Norway, Department of Mapping Sciences, {oystein.andersen|barbi.nilsen} @ikf.nlh.no
Working Group: III/1
KEY WORDS: Integrated sensor orientation, aerial triangulation, GPS, IMU, international test.
ABSTRACT
The topic of image orientation by combined aerial triangulation with GPS/IMU, also called integrated sensor orienta-
tion, has received much attention lately. One of the main questions of fundamental relevance is, if and under which
conditions the direct determination of the parameters of exterior orientation via GPS and IMU can be a complete sub-
stitute for aerial triangulation. A more practical question deals with the possibilities of an optimum combination of the
different methods using a minimum of ground control points. The European Organisation for Experimental Photo-
grammetric Research (OEEPE) has embarked on a test investigating these issues. The main focus of the test is on the
obtainable accuracy of integrated sensor orientation for large scale topographic mapping as determined at exterior ori-
entation elements and at independent points on the ground. In this paper we describe details of the test which, when
presented at the ISPRS Congress in Amsterdam, is still open for interested participants to join.
1 INTRODUCTION
Image orientation is a key element in any photogrammetric project, since the determination of three-dimensional coor-
dinates from images requires the image orientation to be known. In aerial photogrammetry this task has been exclu-
sively and very successfully solved using aerial triangulation since many decades. Thus, aerial triangulation has become
a key technology and an important cost factor in mapping and GIS. Over the years a number of additional sensors were
used to directly determine at least some exterior orientation parameters, albeit with little success until the advent of GPS
in the late eighties. Today differential kinematic GPS positioning is a standard tool for determining the camera exposure
centres for aerial triangulation. Using the GPS measurements as additional observations in the bundle block adjustment
a geometrically stable block based on tie points alone can be formed, and ground control points (GCP) are essentially
only necessary for calibration, for detecting and eliminating GPS errors such as cycle slips and for reliability purposes
(Ackermann, Schade 1993; Jacobsen 1993; Ackermann 1994; Jacobsen 1997; Andersen, Ackermann 2000). Applica-
tions involving image strips such as highway mapping, however, still need GCP in order to reliably determine the rota-
tion of a plane around the flight axis.
Using gyroscopes, one is able to also determine the rotation elements of the exterior orientation. Gyroscopes and accel-
erometers are the components of an inertial measurement unit (IMU)', the accelerometers provide sensor velocity and
position via integration. Thus, a GPS/IMU sensor combination can in principle overcome the mentioned GPS errors and
can yield the exterior orientation elements of each image without aerial triangulation. This technology opens up many
new applications for photogrammetry and remote sensing (Schwarz et al. 1993; Colomina 1999; Skaloud 1999). Ac-
cording to the first author four areas can be distinguished (Schwarz 1998), namely (1) topographic mapping in which
very high accuracy requirements are to be met and standard photogrammetric film cameras are being used today and
will continue to be used for some time to come, (2) applications using the emerging digital aerial line and frame cam-
eras, (3) applications with less stringent accuracy requirements such as mobile mapping enabling the use of less expen-
sive sensors in the air or on the ground, and (4) navigation applications which require a real-time response.
A series of tests and pilot projects have been conducted and have convincingly shown the great potential of GPS/IMU
sensor integration in aerial photogrammetry (Skaloud, Schwarz 1998; Wewel et al. 1998; Abdullah, Tuttle 1999; Bur-
man 1999; Colomina 1999; Cramer 1999; Toth 1999; Jacobsen 2000). At independent check points on the ground root
mean square errors of down to 0.1 to 0.2 m were obtained. These results have proven that the direct determination of the
exterior orientation elements is a significant component of highly accurate image orientation and is a serious alternative
to aerial triangulation in a number of applications. Also, a number of potential error sources have been identified. These
include the Kalman filtering of the GPS/IMU data for noise reduction, the determination of parameters for systematic
position and attitude corrections of the GPS/IMU data, the stability of these parameters over time, especially the stabil-
ity of the attitude values between the IMU and the camera, the time synchronisation between the various sensors, issues
related to the correlation between the interior and the exterior orientation parameters of the imagery, and the quality of
the resulting exterior orientation parameters for subsequent stereoscopic plotting.
! Note, that we use the term IMU instead of INS (Inertial navigation system). Following Colomina (1999), an INS box
contains an IMU as a measurement device plus positioning and guidance functions, mainly realized in software.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 373