Check Point Residuals Parallax 
dX (m) dY (m) dZ (m) (um) 
Min -0.26 -0.29 -0.12 32 
Max 0.09 0.09 0.30 30.4 
Mean -0.05 -0.07 0.12 12.7 
RMS 0.10 0.13 0.15 14.8 
  
  
  
  
  
  
Table 6. EO Analysis Results for the POS AV 510 Data 
     
  
     
     
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
of the boresight mis-alignment error in the simulated 310 data, 
the refined EO parameters is brought to ISAT and another EO 
analysis was performed. This is mainly to validate the simulated 
data with the error budget analysis by comparing the EO 
analysis against the POS AV 510 system. The result is shown in 
Table 9. 
  
  
  
   
    
  
   
   
    
   
    
    
   
   
   
  
   
    
   
   
  
  
    
   
  
    
     
   
     
   
    
  
  
  
    
  
   
  
  
  
  
  
   
     
  
  
  
  
  
  
  
  
  
Check Point Residuals Parallax 
Ensemble dX (m) dY (m) dZ (m) (um) 
RMS 0.24 0.40 0.28 63.5 
  
  
  
  
  
Table 7. EO Analysis Results for the Simulated POS AV 310 
Data (Monte Carlo Analysis) 
It is clear form Table 6 that the ground accuracy of a POS AV 
510 system is very accurate. The 3D position is < 15 cm and the 
parallax is less than | pixel. In contrast, the simulated 310 data 
shows a higher check point RMS, and the parallax is quite 
significant, which is expected for a lower accuracy system. To 
validate these results, they were compared with the error budget 
analysis performed by Applanix Corporation (Mostafa et al, 
2001). In this report, the horizontal ground position accuracy of 
a POS AV 310 was about 2 times poorer than that of 510 
system, while the vertical accuracy is about 1.5 times poorer 
(both for the mapping scale of 1:6000). Although the actual 
differences are slightly higher than the theoretical values, the 
error budget analysis is ideal and assumes an error free system 
other than the error in the direct EO. In the simulated 310 data, 
the check point RMS include errors such as residual boresight 
error, check point accuracy and measurement noise. Hence the 
ensemble RMS derived by the Monte Carlo Analysis can be 
considered as representative of the performance of a true POS 
AV 310 system. The performance of the simulated 310 data is 
summarized in Table 8, which has very similar performance as 
a POS AV 310 system specifications listed previously in Table 
> 
J. 
Check Point Residuals Parallax | 
dX (m) dY (m) dZ (m) (um) 
Min -0.67 -0.47 -0.39 r1 
Max 0.63 0.50 0.80 85.6 
Mean -0.06 -0.03 0.13 380 | 
RMS 0.22 0.27 0.27 402. 
  
  
  
  
  
  
  
  
  
  
  
Post-Processed Accuracy Relative Value 
Position (m) 0.06 — 0.3 
Roll (deg) 0.013 
Pitch (deg) 0.013 
True Heading (deg) 0.036 
  
  
  
  
Table 8. Specification of the Simulated 310 Data 
4.4 Quality Assurance / Control of Simulated Data 
A reference boresight calibration has been performed on the 
POS AV 510 data after the Quality Control / Assurance 
procedure. Since the simulated 310 data shared the same 
hardware as the 510 data, and only the raw IMU data has been 
degraded, the boresight angle of the simulated 310 should be 
the same as reference. But, in order to understand the behaviour 
of lower accuracy system, a QC/QA procedure will be 
performed in the simulated data. Since the Monte Carlo 
Analysis creates 10 sets of simulated data, only one set of data 
will be picked for the rest of the tests in this paper. The 
selection of such dataset is based on the ensemble RMS 
difference and the ensemble check point RMS from EO analysis. 
Georeferencing EO Analysis Test on 
After removal 
4.4.1 Direct 
QA/QC’ Simulated POS AV 310 Data: 
Table 9. Results of EO Analysis on QA/QC'd Simulated POS 
AV 310 Data 
From Table 9, and by comparison to Table 7, it can be seen that 
horizontal accuracy has been improved with the boresight mis- 
alignment error being minimized. Such improvement has 
brought the simulated data closer to the ideal ratio difference in 
the error budget analysis. This again validates the simulated 
data that can represent the POS AV 310 performance. 
5. INTEGRATED SENSOR ORIENTATION TEST 
After validating the simulated data and proper QA/QC 
procedure, further test can be carried on to analysis the 
performance of low cost GPS/INS system using Integrated 
Sensor Orientation technique. In addition to the simulated data, 
the 1SO test will also focus on applying assisted triangulation 
on POS AV 510 data, which is a high performance DGPS/INS 
system. Accuracy improvement of such data will be reviewed 
and compared with the performance when ISO is applied on the 
simulated data. The assisted-triangulation is performed in ISAT 
after automatic tie point collection is performed. Notice that no 
ground control point is used in the assisted-triangulation. 
5.1 ISO Test on Simulated Data 
The test will begin with the simulated data. The test will focus 
on the assisted triangulation result and compare the result with 
the reference data, the POS AV 510 Direct Georeferencing data, 
as shown in Table 6. Starting with using | strip of data, Figure | 
presents the check point RMS when different point per von 
grubber (PPVG) values are used. 
Integrated Sensor Orientation using 1 Strip of Simulated Data 
0.3 
0.25 * oa mmm min ii a a 
E 
o 0.2 
= 
X |(3 Easting 
€ 045 . [8 Northing’ 
© (9 Vertical 
x 
o 
o 
£ 
o 
  
  
2 3 4 5 6 7 8 9 
Point Per Von Gruber 
Figure 1. ISO Test Results when using 1 Strip of Simulated 
POS AV 310 Data