combining all images, selected images can also be used. The 
two lower curves in Figure (10b) show this situation for two 
pairs of images. In general this will be sufficient to achieve 
the desired accuracy. 
As mentioned above, the large errors in the results from 
individual pairs of images are most likely due to an 
: a . 3 c . 
instability of the camera mount, i.e. in a dR,- matrix that 
changes with time. This can be either caused by differential 
vibrations between camera and INS or by physical 
movement of the complete camera mount. The latter is most 
likely the case. In this specific test, a different vehicle was 
used and it was detected later on that the mounting was much 
less stable than usually. 
  
Target Point Figure (10a) 
So 
D 
  
   
  
  
  
  
  
  
  
   
Individual Stereo 
ON 
© 
     
© 
Figure (10b) 
    
mbining pair (2) 
    
Error (cm) 
N Wo B a 
© x © 
   
with all others 
-Ó 
© 
10 20 30 40 
Distance (m) 
Figure 10 : Effect of Geometrical Constraints 
on 3-D coordinates 
6. CONCLUSIONS 
The prototype of the VISAT system presented in this paper 
is the first mobile GIS data acquisition system of this 
accuracy class. It is designed to work with comparable 
accuracy in rural as well as in urban areas where GPS may not 
be available all the time. To maintain a consistent accuracy 
of 0.30 m (RMS), the integration and mutual control of INS 
and GPS is an essential aspect of the system concept. The 
complementary features of INS and GPS permit the 
resolution of cycle slips and outages, as well as INS position 
and attitude control. The tight integration and 
synchronization of the GPS/INS component with the CCD 
camera cluster results in the precise georeferencing of each 
image, which is a necessity for a system of this accuracy 
class and gives considerable flexibility in data processing. 
A major focus of the paper is the analysis of recent field 
tests, designed to estimate the contribution of each 
individual error source to the total error budget. 
Georeferencing can be done with a position accuracy of 5-10 
cm (RMS) for the individual exposure station and an attitude 
accuracy of 10-20 arc seconds for pitch and roll and 30-50 
arc seconds for azimuth. The position results were obtained 
from road tests, the attitude results from lab tests for angular 
velocities of up to 30 degrees per second. The imaging 
component, which provides positioning with respect to the 
current van position, achieves an accuracy of 10-15 cm for 
distances of up to 50 m. Further improvements can be 
expected with a third camera being added to the camera 
cluster. By using redundant measurements instead of 
individual image pairs and by applying geometrical 
constraints, results can be considerably improved and meet 
the production system requirements of 0.30 m in all cases 
studied. 
ACKNOWLEDGMENTS 
This paper is a contribution to the development of the 
VISAT system. The system is a joint project between the 
University of Calgary and GEOFIT INC, Laval, Quebec. 
Darren Cosandier is gratefully acknowledged for his support 
during system testing. 
REFERENCES 
[1] Cosandier, D., Chapman, M. A, High Precision 
Target Location for Industrial Metrology, Videometrics, 
SPIE OE/Technology, Boston, November, 1992. 
[2] El-Sheimy, N., Schwarz K.P., Kinematic 
Positioning In Three Dimension Using CCD Technology, 
VNIS93 Conference, Ottawa, October 12-15 1993. 
[3] Li, R, M. A. Chapman, Qian, L., Xin. Y. 
and K. P. Schwarz, Rapid GIS Database Generation 
Using GPS/INS Controlled CCD Images, ISPRS 94 GIS/SIG, 
June 6-10, Ottawa, Canada. 
[4] Schwarz, K.P., D. Lapucha, M.E. Cannon, H. 
Martell, The Use of GPS/INS in a Highway Inventory 
System. Proc. FIG XIX Congress, Helsinki, Finland, Vol. 5, 
paper 508.4, pp. 238-249, May 10-19, 1990. 
[5] Schwarz, K. P., Martell, H., El-Sheimy, N., 
Li, R., Chapman, M., Cosandier, D. (1993a): 
VISAT- A Mobile Highway Survey System of High Accuracy, 
VNIS Conference ‘93 Conference, Ottawa, October 12-15, 
1993. 
[6] Schwarz, K.P., Chapman, M.A, Cannon, M. 
W., Gong, P. (1993b): An Integrated INS/GPS Approach 
to the Georeferencing of Remotely Sensed Data, PE&RS Vol. 
59, No. 11, November 1993, pp. 1667-1674.. 
[7] Wei, M. and K.P. Schwarz (1990a): Testing a 
Decentralized Filter for GPS/INS Integration. Proc. IEEE 
PLANS 1990, Las Vegas, pp. 429-435, March 20-23, 1990, 
[8] Wei, M. and K.P. Schwarz (1990b): A Strapdown 
Inertial Algorithm Using an Earth-Fixed Cartesian Frame. 
Navigation, Vol. 37, No. 2, pp.153-167, 1990. 
248 
  
Int 
en 
(SIG 
quan 
manu 
par 
Afin 
util 
expl 
véhi 
syst 
ou € 
grou 
de 
navi 
des 
de 
exp] 
pos: 
Cor! 
COUJI 
cha« 
L'h 
obti 
nav. 
1l 
néc 
COO 
pro 
On 
Coo 
ess