ul 2004
with
irect
erlal
the
only
natic
t EO
ional
truly
1 for
direct
S are
clude
d the
d by
e last
ation.
un a
direct
it and
ctical
| ISO
JA) /
| part
A/QC
rated
self-
{ one
area.
ined.
| EO
ers.
scale
y of
ently
ping.
area
being
o fly
] ifa
ower
stem
it has
ain a
lirect
terior
there
racy.
o be
ure.
ISO
direct
ower
ts of
t the
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part BS. Istanbul 2004
4. TEST DATA PREPARATION
The ISO investigation was performed using real-flight test data
from a high-end Applanix POS AV 510, and a simulated data
from a less-accurate POS AV 310 for the same flight. To
simulate the performance of the less accurate system, the IMU
data from the POS AV 510 was degraded using statistical error
models based upon Applanix Corporation’s proprietary
simulations tools. This method allows a direct comparison
between two datasets for the same operating conditions:
identical ground coverage, number of photos and flight
trajectories.
4.1 Reference System Description
À RMK Top film camera data equipped with Applanix POS AV
510 system was selected for the test. The data parameters are
listed in Table 1. The published accuracy specifications for the
POS AV 510 system are presented in Table 2.
Noise (deg/sqrt(hr)) 0.15
IMU Drift (deg/hr) 0.5
Table 3. Specification of POS AV 310 System
The primary difference in system performance between POS
AV 310 and POS AV 510 system is the orientation accuracy,
which is directly a function of the IMU. Therefore, the raw 510
IMU data was brought into a simulation tool and purposefully
degraded. The simulation tool superimposes additional random
noise, bias, scale factor, and mis-alignment errors on both the
accelerometer and gyro data. After running through the tool. the
degraded IMU data was then post-processed with the original
unaltered GPS data using the Applanix’s POSPac™ software
(Post-Processing Package). The simulated POS AV 310
solution was then differenced with the original POS AV 510
solution and the RMS differences in both position and attitude
were computed. To statistically validate the simulation, a Monte
Carlo Analysis was performed and ensemble RMS on both
position and attitude difference was determined. Table 4
presents the theoretical RMS value for the differences between
a POS AV 510 and 310 based upon their specifications. If the
simulation is valid, the ensemble RMS of the differences should
approach these values. The ensemble RMS differences results
from the Monte Carlos analysis are presented in Table 5.
Location University of Kentucky,
United States
# of Strips 4
# Photo / Strip 8
Flying Height (m) 900 AGL
Scale 1: 6000
Photo Scan Resolution (um) 15
Forward / Side Overlap 60% / 20%
Mapping Projection
StatePlane Zone 1601
Navigation Parameters
Ideal RMS Difference
Position (m)
0
Roll & Pitch (arc minute)
0.72
Datum, Height WGS84, Orthometric
# of Check Points 18
DGPS/INS System Applanix POS AV 510
Table I. Dataset Information
| Post-Processed Accuracy Absolute Value
Position (m) 0.05 — 0.3
Roll & Pitch (deg) 0.005
True Heading (deg) 0.008
Noise (deg/sqrt(hr)) 0.02
IMU Drift (deg/hr) 0.1
Table 2. Specification of POS AV 510 System
To simulate as much as possible a perfect system calibration,
any residual boresight errors were removed using the Quality
Control/Quality Assurance procedure documented by Applanix
Corporation with its POSCal™ (IMU/Camera Calibration
Software). The original camera calibration from U.S.G.S. is
used and assumed to be correct.
42 Less Accurate Direct Georeferencing Data Simulation
To investigate the performance of ISO using a lower accuracy
DGPS/IMU system compared to a high accuracy POS AV 510
system, the POS AV 510 data was degraded to simulate the
performance of a POS AV 310 system. A POS AV 310 system
was chosen since it achieves approximately 3 times lower
orientation performance than the 510 system. Table 3 presents
the published specifications of the POS AV 310.
Post-Processed Accuracy Absolute Value
Position (m) 0.05 — 0.3
Roll & Pitch (deg) 0.013
True Heading (deg) 0.035
Heading (arc minute) 2.04
Table 4. Ideal RMS Difference
Navigation Parameters Ensemble RMS Difference
Northing (cm) 2.85
Easting (cm) 2.40
Vertical (cm) 1.37
Roll (arc minute) 0.74
Pitch (arc minute) 0.71
Heading (arc minute) 2.08
Table 5. Ensemble RMS difference of the Simulated 310 Data
From Table 5, the ensemble RMS difference is very close to the
ideal RMS difference given in Table 4, which validates the
simulation. However, in order to further validate the simulation.
each Monte Carlo trial was analysed using the EO Analysis tool
from the Z/I ImageStation Automatic Triangulation software
(ISAT).
4.3 Direct Georeferencing EO Analysis Test
This test is performed to validate the simulated data in addition
to the ensemble RMS difference obtained from the Monte Carlo
Analysis. The EO Analysis evaluates the quality of exterior
orientation parameters by comparing the given coordinates of
check points with the intersection of the rays of these points as
project it on the overlapping photo pairs by the EO Data. Table
6 lists the EO analysis result for the POS AV 510 data, while,
Table 7 lists the ensemble RMS from the Monte Carlo Analysis
of the degraded data. Notice that the EO Analysis Results
presents the statistics of check point residuals for all check
points used in the EO Analysis.
