International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
For precision applications, however, the navigation data is 
either processed using a single dedicated GPS basestation or 
multiple base stations such as the CORS (c.f., Snay, 2000). In 
post-mission, the precise navigation data together with the 
digital images go through calibration and quality control for 
precision mapping purposes, where a number of accurate and 
well distributed tie points are generated in fully or semi- 
automatic mode, which are then used to perform a calibration 
and quality control procedure to refine the boresight, camera, 
and datum calibration parameters, as shown in Figure 2 
2. TERRESTRIAL CALIBRATION RESULTS AND 
ANALYSIS 
The geometric accuracy of the mapping products produced 
using the DSS data depends on the resolution and the accuracy 
of each single component of the entire system and the accuracy 
of the calibration parameters of the system components and the 
system as a whole. This section is dedicated to present the 
results of the terrestrial calibration of the DSS system. 
{mage upper " 
left corner y Y i 
> x? 
  
p A 
X pps rac 
/ Y ppac X : 
/ » 
Principal Point of i 
Autocollimation Image centre 
(PPAC) 
  
  
  
  
4092 
Lines/Rows 
  
  
te 
  
MM MM rn 49 PielsiCcoliums 
  
Remarks 
|l. / X'and Y *: Image Coordinate Frame — Right Handed System 
  
mitor Coordinate Frame ~ Lelt Handed System 
Figure 3: The DSS Image Coordinate System 
  
Radial Lens Distortion (um) 
$5293 4^ 80 9..7 (8 9:10 14.12 3. 1. 15. 18 17/48 19 20 21 22 23 24 25.26 
Radial Distance (mm) 
-*- SN0028 -8&- SN0032 —- SN0033 —- SN0036 ~~ SN0037 -G- SN0041 -- SNO035 
Figure 4: Radial Lens Distortion Profile for Different DSS 
Systems — 55 mm lenses 
The terrestrial calibration is done using traditional digital 
camera calibration techniques (c.f., Beyer, 1992, Fraser, 1997, 
Lichti and Chapman, 1997) using a target field consisting of 
over 160 well surveyed targets with a sub-millimetre surveying 
precision surveyed by a Total Station. This target field is 
imaged by the DSS system in close-range from different 
surveyed locations and with different orientation angles to 
allow for precise calibration of the focal length, principal point 
offsets, and lens distortion, as well as an initial value of the 
boresight parameters. For details, see Mostafa, 2003 for the 
DSS and Mostafa and Schwarz 2000 and Mostafa and Schwarz, 
2001 for a detailed description of the overall system calibration. 
^4 
As shown in Figure 3, the position of the principal point of 
autocollimation is computed in the image coordinate frame and 
the boresight angles rotating that frame into the IMU frame are 
also computed during the DSS terrestrial calibration. Although 
the radial lens distortion of the lenses is normally large, as 
shown in Figure 4, it is calibrated in the terrestrial mode and 
compensated for in the photogrammetric reconstruction. of 
image rays during the map compilation mode. 
Figure 5 shows the repeatability of the radial lens distortion 
profiles using different lenses (shown in Figure 4) when 
compared to the first lens profile. It is noticeable that the 
difference between the radial distortions of totally different 
lenses is well within one pixel (9 um). This seems to indicate 
that the lenses used by the DSS possess almost similar radial 
distortion character and implies that these lenses can be 
swapped in the field which will result in a positional error 
contribution of no more than one pixel at the image edge, 
without recalibrating the system. However, this is not 
recommended to be done in the normal practice since there is 
no guarantee that this conclusion is consistent and it is always 
best to calibrate the lens in the terrestrial mode where the 
geometry of the calibration data is a lot more controlled than 
that of the airborne environment. 
  
Radial Lens Distortion (um) 
123456 7 8 9 1011 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 
Radial Distance (mm) 
— SN0032 —- SN0033 -e- SN0035 — SN0036 ——- SN0037 -€- SN0041 
Figure 5: Repeatability of Radial Lens Distortion for Different 
DSS Systems 
   
i ] 
E | | 
2 | 
= 
ge | | 
iic 
9 | 
ES | 
£5 | a 
QE UIT 
a C 
ox 4 i 
2e 
Se 
= E + i 
ES #1 | 
£ | ! 
= | | 
a | 
40 L- ——— 
no, filtor VIS1, No2 vis2 
Calibration ID 
BdXpp OdYpp 
Figure 6: Principal Point Difference Using Different Filters 
Figure 6 shows the difference in the calibrated principal point 
offsets where a lens filter is removed (no filter) and then placed 
back on the lens (VIS1 No2) and totally replaced by a new 
filter (VIS2). When the lens filter is placed back on the lens and 
the system is re-calibrated, the difference of the principal point 
offsets is no more than one micron. On the other hand, when no 
filter is used or a totally new filter is used, the difference in the 
calibrated principal point offsets is well within one pixel, which 
is easily calibrate during the airborne quality control procedure 
discussed later. This indicates that the DSS user can replace 
lens filters without the need for recalibrating the DSS system in 
the case of the Rapid Response applications. It is recommended, 
however, to calibrate each lens together with its filter in the 
terrestrial mode for precision mapping applications. 
     
   
  
    
   
  
  
  
    
    
   
   
   
   
   
   
   
    
   
   
    
     
     
   
   
    
       
   
   
   
    
    
    
   
   
  
    
   
   
  
"-——— PR ^ nd rt amd ar qt mt A 
"— —— Rm 4 t