Table 1 : Error sources of the VISAT system and their magnitude 
  
Error in Expected Magnitude 
. : m 
Contribution to or; Characteristics 
  
* First term : 
= ris (INS/GPS). 
  
10 - 20 cm 10 - 20 cm Constant for all points in the 
same image. 
2- Synchronization 1 - 3 milliseconds 2-6cm 
* Second term 1- Function of the distance 
m between the camera and the 
1 OR, (D (INS) l arcmin 1-2cm target. 
2- Synchronization 0.5 arcmin 1 cm 2- Constant for all points in 
the same image. 
  
  
* Third term 
b by i. 
1- dR (Calibration) 10 - 20 arcsec 
2- sa? (Calibration) 
1-2 cm 
3- Sr (Target pointing) One pixel 
  
  
1- constant for the whole 
maximum 1 cm 
mission. 
1-2cm 2- constant for the whole 
mission. 
1-8cm 3- Function of the pixel size. 
  
  
  
  
Therefore, all time tags can be put in the GPS time frame 
every time a call is made to the PC timer. One problem with 
this technique, is the synchronization of the CCD cameras. 
The problem arises from the fact that most of the CCD 
cameras do not issue an interrupt indicating the time of 
exposure. Therefore, it should be synchronized 
mathematically using the specifications of the Video rate. 
The maximum error in synchronizing the cameras is of the 
order of 3.0 milliseconds as shown in figure (5). Results in 
this figure were calculated using the following equation: 
Synchronization error = 2 - (t1 + 1/30) 
where: 
tl is the computer time before grabbing the images; 
t2 is the computer time after grabbing the images; 
1/30 is the time of grabbing one image. 
  
  
o 
S 
3 
  
-0.002 
nchronization Errors (sec) 
  
à? -0.004 2 
  
  
  
-0.006 i i L 1 A 1B , i 1 1 
57435 57445 57455 57465 
Computer Time (sec) 
  
Figure 5 : Camera Synchronization Errors. 
The GPS contributes to the georeferencing process by being 
the main positioning component for the perspective center 
m ; os 
vectors Foc) of the two cameras. For precise position 
determination, differential GPS is needed. A variety of 
differential kinematic truck surveys using GPS have been 
performed over the last few years. Most of these tests have 
shown some limitation due to low data rate, outages due to 
satellite masking which cause poor geometry, and loss of 
carrier phase lock. Therefore, the use of GPS position and 
velocity will be limited to updating the INS data stream. The 
motivation for the integration of INS and GPS is to exploit 
the benefits of each system. The achievable accuracy of 
INS/GPS for many road tests is at the level of a few 
centimeters when a consistent satellite constellation is 
available. Typical INS/GPS positional errors for a truck 
survey over a baseline of 16 km is shown in figure (6). 
Agreement with pre-surveyed control points was generally 
better than 10 cm with an RMS of 7 cm. 
  
  
  
  
  
  
  
  
  
40 
E North 
re Fast 
30 P " 
$ I 
X 20 
10 X a 
$ = B Bg 
a 
ol eese 
4 6 8 10 12 14 16 
Distance (Km) 
Figure 6: GPS/INS Positional Errors. 
The INS contributes to the georeferencing process by 
determining the position and attitude of the c-frame, i.e. by 
m ; : b 
giving the vector re and the attitude matrix R ml). For 
attitude determination, the INS is the primary system. In 
order to reach the required accuracy from the VISAT system, 
the derived INS roll, pitch, and azimuth should be at the level 
of one arcminute. To check such accuracies, lab tests were 
conducted at the U of C Mechanical Engineering Department. 
246 
In this t 
platform 
of 3 arcs: 
attitude 
platform. 
errors are 
and 30-5 
or less f 
confirm © 
> > Rn | 
o 239 s. € 
q— 7 a er > 1 
  
Angular Accuracy (arcsec) 
9 
d. m 
The up 
interpol 
slips. 1 
accurac: 
these t 
interpo. 
positior 
typical 
static m 
latitude 
second, 
millime 
interva 
0.0: 
= 00 
Latitude Errors (m 
-0.0 
The id 
measu: 
from t 
respec 
intege: 
value. 
interv: 
for ins 
depen 
correc 
be acc 
The «