in, K., Hjort, N.L., 
lontextual Classific- 
istical Methods and 
Report 768, Norwe- 
a 1996 
EXTERIOR ORIENTATION BY DIRECT MEASUREMENT 
OF CAMERA POSITION AND ATTITUDE 
J. Skaloud, M. Cramer* and K.P. Schwarz 
Department of Geomatics Engineering 
The University of Calgary 
2500 University Drive NW, Calgary 
Canada (T2N 1N4) 
e-mail: skaloud@acs.ucalgary.ca 
* Institute for Photogrammetry 
University of Stuttgart 
Keplerstr. 11, 70174 Stuttgart 
Germany 
e-mail: michael.cramer@ifp.uni-stuttgart.de 
ISPRS Commission III, Working Group 1 
KEY WORDS: Photogrammetry, Integration, Sensor Fusion, Integrated GPS/INS-System, Direct Exterior Orientation. 
ABSTRACT 
Airborne attitude and position determination as a means of determining the exterior orientation of an airborne remote sensing system is 
investigated in this paper. The performance of an airborne data acquisition system consisting of receivers of the Global Positioning System 
(GPS) and a strapdown Inertial Navigation System (INS) together with an aerial camera is assessed using data of a 1:6000 large scale 
photogrammetric test. The test was jointly conducted by the Institute for Photogrammetry, Stuttgart and The Department of Geomatics 
Engineering, Calgary, with aircraft and logistics support by Rheinbraun AG - Department of Photogrammetry, Cologne. Multiple flight lines 
were flown over a well controlled photogrammetric test field allowing the assessment of position and attitude repeatability, as well as the 
analysis of gyro drift. 
After a brief description of the essential features of the sensor integration design, its practical implementation is described and the error budget 
of the GPS/INS integration is discussed. The actual position and attitude results obtained from the GPS/INS are then compared to those derived 
from the independent aerotriangulation bundle adjustment using all available control points. The errors in position determination along the 
aircraft trajectory are in the decimetre range, those in attitude are varying with standard deviation of 0.03 degree over one hour. To assess the 
feasibility of using independently determined attitude and position parameters from GPS/INS for the exterior orientation of the photographs, 
the independent models were directly georeferenced. Preliminary results indicate that aerotriangulation at a photo scale of 1:6000 using 
independent exterior orientation directly obtained from an integrated GPS/INS can be done with an accuracy of 0.3 m (RMS). It is expected 
that these results can be improved because there are considerable doubts about the accuracy of the synchronisation between the camera and 
the GPS/INS in this specific flight. 
1. INTRODUCTION 
The problem of georeferencing images of aerial photography can be 
defined as the problem of transforming the image coordinates in the 
camera frame to the mapping frame. Such a transformation can be 
written as 
X, X, x^ 
p E p 
Y, EY + aR; (w,9,¥) y, (D 
Z, x 2, » -f R 
where (X,, Y,, Z, ) and x", 3," are the point coordinates in the 
geodetic reference system and the reduced image coordinates in the 
photo frame, respectively; (X 0, Y 9, Z9) are the spatial coordinates 
of camera perspective centre given in the reference frame; & isa 
point dependent scale factor; fis the camera focal length and R;" is 
three-dimensional transformation matrix which rotates the photo 
frame into the geodetic mapping frame. In this equation, the vector 
(X 9, Y o, Zo) and the transformation matrix £5" are time-variable 
quantities. 
In order to georeference frame based imagery, the parameters of 
interior and exterior orientation have to be determined. The interior 
orientation parameters, i.e. coordinates of the principal point x, y ;. 
the focal length f; and the geometric distortion characteristics of the 
lens, can be measured via laboratory calibration. The six parameters 
of camera exterior orientation (X 9, Yo, Zo, & ©, X) are found by 
correlation between ground control points and their corresponding 
images. Such a process will be called inverse photogrammetry. 
Within this method the image coordinates of known control points 
are measured and related to the ground assuming a perspective 
projection. Connection between multiple images is formed by 
measuring points common to adjacent images and by enforcing 
intersection constraints between them. To be able to resolve the 
parameters of exterior orientation and to control the error 
propagation, ground control points have to be established for each 
block of images. This represents a significant portion of the 
aerotriangulation budget. Additionally, the evaluation of the images 
is very time consuming and highly skilled operators are necessary. 
Moreover, the cost of determining ground control points can be 
prohibitive for image georeferencing in remote areas. 
If the parameters of exterior orientation can be derived from 
simultaneously flown on-board sensors with sufficient accuracy, the 
number of ground control points can be reduced, resulting in obvious 
economic advantages. The potential of using GPS observations as 
constraints for the camera perspective centres in bundle adjustment 
has been proven repeatedly (Friess 1991, Ackermann 1995) and 
GPS-supported aerial triangulation is by now an accepted procedure. 
In case of pushbroom imagery, parameters of exterior orientation are 
required for each scan line. Applying a block adjustment procedure 
to this problem would require very large numbers of control points. 
Several rather complicated solutions have been proposed to 
overcome this problem (Hofmann 1988, Hofmann et al. 1993), but 
none of them has been accepted in practice. Again, a direct solution 
125 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996