The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
90
GPS, IMU, etc, is of high importance, since sensor orientation
is very sensitive to changes in sensor geometry because of the
extrapolation from the camera projection center to the ground.
The camera calibration process involves the geometric
calibration, resolution determination and radiometric calibration
of camera. The laboratory calibration is a standard method for
analog airborne frame cameras and the interior camera
geometry, i.e. focal length, principle point location and lens
distortion parameters are estimated. In-situ calibration, also
called self-calibration originates from close range applications
and requires a large number of signalized control points in 3D
terrestrial calibration fields. The airborne in-situ calibration also
requires a calibration field with signalized control points of
high accuracy. The focal length, principle point location and
lens distortion parameters are estimated during the calibration
process from the measurements reflecting actual conditions.
The ASPRS and the EUROSDR continue their research on
digital camera calibration and validation (Stensaas, 2005,
Cramer, 2005). The lab calibration method is still used for
geometric calibration of large format digital cameras such as ZI
Imaging DMC and Leica ADS40 together with in-situ
calibration (Dorstel et al., 2003, Tempelmann, et al., 2003). For
medium format digital cameras such as the Applanix/Emerge
DSS geometric calibration is done by terrestrial and airborne
calibration (Mostafa, 2004). The terrestrial in-situ calibration
method is used for geometric calibration of the Vexcel
Ultracam D large format digital camera (Kropfl, et al, 2004). The
U.S. Geological Survey (USGS) began its certification efforts
with digital aerial cameras in January 2006 (Stensaas, 2006).
Current investigations at USGS focus on geometric and spatial
characterization and calibration, although research on
radiometric characterization and calibration of airborne sensors
also continues with the recently established digital camera
calibration lab at the USGS Center for Earth Resources
Observation and Science (EROS). To perform in-situ camera
calibration, a calibration range was established also at USGS
EROS with 150 surveyed panels and one CORS station
(http://calval.cr.usgs.gov).
The offset between the GPS antenna and the IMU or imaging
sensor can be precisely measured by conventional survey
methods. The boresight misalignment is determined by a
comparison of the GPS/IMU derived image orientation with the
results of a bundle block adjustment over a calibration field
containing control points. Finding such a calibration field
nearby or within project area, however, is not always possible.
This research was motivated by the circumstances of a LiDAR
mapping project, where a small format digital camera was
installed next to the LiDAR system in the last minute before the
airborne surveys started. There were no measurements and no
dedicated data acquisition performed to support boresight
calibration of the camera except for an earlier lab camera
calibration. Furthermore, there was no target range or any area
with signalized controls in the project area. Therefore, this
investigation aimed to determine the system calibration
parameters for that multi sensor airborne system using only the
LiDAR data acquired in the project. The multi sensor airborne
system included the small format Redlake MS 4100 RGB/CIR
digital camera and the Obtech 3100 ALSM, supported by the
Applanix georeferencing system (Novatel GPS and LN200
IMU). Experiments with the data set and the results obtained
are analyzed and discussed to assess the performance of system
calibration for the Redlake MS 4100 camera.
2. PROJECT AREA AND DATA ACQISITION
The data set from the project B4 Airborne Laser Swath
Mapping (ALSM) survey of the San Andreas Fault (SAF)
System of Central and Southern California, including the
Banning Segment of the SAF and the San Jacinto Fault system
was used (Toth et al., 2007). The project B4, codenamed to
reference to the “before” status of a widely anticipated major
earthquake, the Big One, is a National Science Foundation
(NSF) sponsored project, led by scientists from The Ohio State
University (OSU) and the U.S. Geological Surveys (USGS), to
create an accurate surface model (DEM) along the San Andreas
and San Jacinto Faults in southern California. Besides the
USGS, the OSU-led team included NCALM (National Center
for Airborne Laser Mapping) from the University of Florida,
UNAVCO (a non-profit, membership-governed consortium,
supporting Earth science) and Optech International.
The airborne surveys took place May 15-25, 2005 with a
Cessna 310 aircraft which was hired and Optech International
provided the ALTM 3100 system. The state-of-the-art Optech
3100 system was configured for 70 kHz, resulting in an about 2
pts/m 2 LiDAR data. An experimental Redlake MS 4100 digital
camera was installed next to the Optech 3100 system, providing
imagery of 1K by 2K resolution in four bands. The images were
captured at 1 Hz, synchronized to the 1PPS GPS signal. The
project area, encompassing about 1,000 km of fault line, was
segmented into smaller sections during flight planning,
including the San Andreas and San Jacinto Fault lines.
3. GEOMETRIC CALIBRATION
An experimental digital camera used in this project was the
Redlake MS 4100 RGB/CIR digital camera, which has four
CCD sensors each with 1920 x 1075 pixels and pixel size of 7.4
pm x 7.4 pm. The laboratory camera calibration was performed
in RGB mode and a focal length was estimated for 25.966 mm
in the calibration report. The calibration values were used as
initial calibration parameters for the in situ calibration of the
Redlake MS 4100 camera. The camera was configured for CIR
image capture for the airborne surveys. The CIR imagery has
1892 x 1060 pixels, different from the RGB images. The
approximate image size is 8 mm in the flight direction and 14.2
mm across the flight direction.
A reference block was selected from a typical flight on May 25,
2005 with 21 images, where several man-made structures were
available (Figure 1). 25 control points were used to perform an
airborne in-situ calibration of the Redlake MS 4100 digital
camera which were extracted from the LiDAR point cloud and
intensity data (Figure 2).
Figure 1. The 3D view of geo-registered LiDAR point clouds