recursive
anuscripta
Biases of
'0]. First
cations of
ning, The
essing for
45, Delft,
nbul 2004
THE EFFECT OF SYSTEM CALIBRATION ON DIRECT SENSOR ORIENTATION
N. Yastikli
Yildiz Technical University, Civil Engineering Faculty, TR-34349 Besiktas, Istanbul, Turkey
ynaci@yildiz.edu.tr
KEY WORDS: Photogrammetry, GPS, IMU, Direct, Sensor, Orientation
ABSTRACT
The determination of image orientation parameters of any sensor during data acquisition became possible by combined use of an
inertial measurement system (IMU) and GPS. In this integrated system, GPS antenna, IMU and imaging sensor are located different
position in airborne carrier. Because of this reason, the displacement vectors between sensors have to be determined. Similarly, axes
of the IMU and imaging sensor are not same and a mis-orientation matrix exists between them. System calibration is including both
calibration of individual sensor and calibration between sensors. The IMU calibration for drifts and biases and the calibration of
imaging sensor for interior orientation parameter are components of sensor calibration. Calibration between sensors contains the
determination of a constant displacement vector between sensors and a constant mis-orientation matrix between IMU body frame
and imaging sensor frame. The boresight misalignment, the relation between the IMU and the imaging sensor is determined by
bundle block adjustment using a calibration flight. The small change of correction of interior orientation and 3 shifts and 3
misalignment angles between IMU and imaging sensor directly affect direct sensor orientation.
In this study, the effects of the system calibration on direct sensor orientation is investigated based on data set of the test ‘Integrated
Sensor Orientation? of the European Organization for Experimental Photogrammetric Research. For this, bundle block adjustments
have been done with different approach using calibration flights. Using these bundle block adjustments, correction for interior
orientation and 3 shifts and 3 misalignment angles between IMU and imaging sensor have been determined. The object coordinates
of measured image points have been intersected based on GPS/IMU data improved by the boresight misalignment. For each
approach, computed checkpoints coordinates have been compared with given reference coordinates. The effect of system calibration
on direct sensor orientation has been analyzed comparing results of different georeferencing results using different system
calibration parameters.
1. INTRODUCTION
The image orientation is a key element for any kind of imagery
from terrestrial, airborne or satellite based sensors.
Traditionally, this task is solved indirectly by using bundle
block adjustment in photogrammetry. Today, GPS supported
aerial triangulation is successfully used many photogrammetric
map production projects. For new sensors such as LIDAR, SAR
sensor and CCD line cameras, this indirect method can not be
used because of the requirement of exterior orientation
parameters for each scan line (Schwardz at all, 1993). The
direct measurement of image orientations during image
acquisition is appropriate solutions for these sensors using
GPS/IMU system. By combined use of GPS and IMU (Inertial
Measurement Unit), the direct measurement of exterior
orientation parameters (Xo, Yo, Zo and o, q, x) of any sensor
became possible. The GPS/IMU integrated system also can be
used traditional field where bundle block adjustment is used.
The georeferencing of images, recorded by different sensor can
be defined as a transformation problem. For traditional
photogrammetric cameras, this problem cover transformation
form the image coordinates in camera coordinate frame to the
mapping frame. To the georeferencing aerial images, the
interior and exterior orientation parameters have to be
determined. In this context, direct sensor orientation can be
described as the determination of the sensor orientation
parameters based on GPS/IMU data respecting the
(e.g. sensor calibration). Based on the direct georeferencing,
c
determination of the geometric information of the used sensor
object coordinates corresponding to measured image points are
determined.
The direct sensor orientation is based on GPS and IMU data
integrated with Kalman Filter. Three orthogonal mounted
gyroscopes and three accelerometers are the components of an
IMU. In some publication, the term inertial navigation system
(INS) is used instead of IMU. INS contains an IMU as a
measurement device as well as positioning and guidance
functions (Colomina, 1999).
Inertial navigation systems were at first developed for military
navigation applications in 1968. During the 1970s, the
surveying community realised that INS or GPS/INS integrated
system can be used as a survey instrument. In the late 1980s
and early 1990s experimental studies have been done by the
Ohio State University and the University of Calgary (for
details, see Scherzinger B. M., 2001). In resent years, a series of
tests, pilot projects and several publications confirmed the
accuracy performance of direct georeferencing and integrated
sensor orientations (Schwarz at al, 1993; Schwarz, 1995;
Skaloud et al., 1996; Jacobsen, 1999; Colomina, 1999; Cramer,
1999; Skaloud 1999; Heipke et al., 2001; Mostafa and Schwarz,
2001).
In the following, the effect of system calibration on direct
sensor orientation and not correct data handling are investigated
based on the data set of the test ‘Integrated Sensor Orientation’
of the European Organization for Experimental
Photogrammetric Research (OEEPE).