2004 
ection 
ps are 
vious 
nates. 
as an 
inputs 
dom- 
ption 
a file 
1 later 
ndard 
input 
from 
ading 
truth 
te the 
1S) to 
ment. 
to be 
tes of 
as the 
ulated 
as the 
(fined 
GPS 
hat is 
ed to 
levant 
carity 
n that 
K are 
Ines). 
‘true 
from 
The 
"each 
9 and 
round 
This 
error 
)utine 
undle 
tines) 
ion’ 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B6. Istanbul 2004 
    
Edit View Help 
Create Simulated Data + Bundle Adjustment 
Either leave default settings or enter new values 
  
Photo Parameters Control and Observations 
  
  
  
Number of Phates [197 (2-20) Photo Point Übservation Emo — [5 (microns) 
NumberofStips [3 (1-23) GCP Observation Etror [005 (metes) 
Forward Overlap {50 (20-60) X co-ordinate of First Photo [1000 (metres) 
Side Üverlap DT en Y co-otdinate of First Photo [2000 — (metres) 
Focal Length (0. 1524 (metres! Direction of First Flight Line [90 (degrees) 
Flying Height [1520 (metres) In-flight GF'S - 1sd Position Error [005 (meties) 
Photo Scale m (250 - 50000) In-flight GPS "weight [017 (metres) 
Photo Tilt 5 7 (0-3 degrees) In-flight IMU - 15d Angular Error [moor (degrees) 
Terrain Variation — [26 — (metres) In-flight IML "Weight" [0.005 (degrees) 
  
{ 
Filter GCPs ? | 
i 
| Confim Data : Simulate Mission 
  
Complete Solution | 
  
  
  
Figure 7 — SIMAT?2 Screen Shot — User Data Input 
Subroutine Random: At the start of the simulation process, 
this sub-routines generates an array of 100,000 pseudo-random 
numbers (x) with user-defined mean (m) and standard deviation 
(s) using a "sum of uniform deviates tactic". This data is 
heavily used by the other sub-routines in the generation of 
random errors. 
23 Postgraduate Results 
Image orientation is a key element in any photogrammetric 
project, and this task has been exclusively and very successfully 
solved using traditional aerial triangulation for many years. 
More recently a great deal of theoretical and practical research 
and development has allowed a combination of in-flight GPS 
and IMU observations to directly determine exterior orientation 
parameters and/or provide additional observations in the aerial 
triangulation bundle adjustment. 
During 2002-3, one of the students studying for an MSc in 
Geodetic Surveying at the IESSG carried out a project to 
investigate the quality and accuracy available from traditional 
Arial triangulation, GPS/IMU observations assisted aerial 
triangulation and GPS/IMU direct exterior orientation. As the 
main indicator/metric, residuals for each ground point were 
derived from the discrepancy between the final estimated 
ground coordinates calculated in the triangulation and the initial 
coordinates calculated from the simulation. 
In each series of tests, four simulated sets of observation data 
were used: image point observations, ground control point 
observations, GPS antenna phase centre position observations 
and IMU observations. The importance of ground control 
points and the quality and (potential) impact of directly 
measured exterior orientation parameters were examined for a 
series of different “flights” including a short, single strip of 
between 5 and 20 photos, and blocks varying in size from 2 
photos by 5, to 3 photos by 20. Within each “flight”, further 
experiments explored the effect of additional observations from 
GPS/IMU on the aerial triangulation solution via three 
configurations of ground control: three ground control points in 
cach photo, 4 ground control points (in the corners of the block 
of photos) and no ground control points. 
105 
The resulting solutions were internally statistically assessed, 
and also compared to real boresight data (including GPS/IMU 
output from an Applanix POS-AV system, ground control data, 
and aerial triangulation solutions) provided courtesy of 
Simmonds Aerofilms, UK. Very pleasingly, the simulation was 
found to accurately model the real world with the simulated 
Simmonds mission producing statistically identical results to 
the actual results. 
The simulation suite is now being further utilised and developed 
by a PhD student as part of their work in the area of “The 
Future of Aerial Triangulation”. 
3 CONCLUSIONS 
The IESSG has developed a number of simulators which have 
not only been used on a number of commercial and research 
projects but are playing an ever greater role as teaching tools at 
undergraduate and masters level. 
In this paper we have presented a brief summary of some of the 
key packages developed and used over the last 10 years as well 
as a more detailed explanation about our more recent aerial 
triangulation / in-flight GPS/IMU simulator (including 
examples and student results). 
Feedback about the use of simulation in teaching has remained 
very positive over the last 10 years, although it is perhaps not 
surprising to note that students consistently prefer access to a 
(full time?!) real person to guide and assist over a computer- 
based simulation. However, the simulation(s) are almost always 
preferred to a textbook-based approach to teaching and 
learning. 
Further details on all of the simulators, facilities and projects 
described in this paper may be obtained by contacting the 
IESSG. 
4 ACKNOWLEDGEMENTS 
We acknowledge the support of the Engineering and Physical 
Sciences Research Council (EPSRC) received via the Masters 
Training Package (MTP) initiative.