International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
Pixel Translational Motion
@ 29.97 frames/s
8.000 .-
7800 —— 25 km/h
E 600 | — 50 km/h
Enon 75 km/h
g 400 100 km/h
$ 3.000 |
& 2900 | — 125 km/h
1.000 — 150 km/h
000) |l——— REL — 175 km/h
0 100 200 300 400 500 — 200 km/h
Altitude, m
Figure 5: Pixel Motion Estimate Due to Speed and Flying
Height
Feature X Pixel
150 160 170 180 190 200
0
50 4
3
& 100
-
2
B 150
ih
200
Figure 6: Example Hot Spot Feature Track
2.3 Real-time georeferencing
Real-time georeferencing is accomplished by using the
WADGPS/INS information to georeference the LWIR images.
The physical relationship between the GPS antenna, the INS,
the camera and the ground target is shown in Figure 7.
MAPPING FRAME
Figure 7: General Geo-Referencing Diagram
= Mee = ;
In Figure 7, Fp indicates the position vector of the hot spot
with respect to the mapping coordinate system. This position
vector can be obtained mathematically as below.
M. .M 3 M bí. b P..c
rp =TFgps (t)- R, (t)R? TN — Hr, ) (1)
Where:
n Is the X, Y, Z position vector of the feature point in
mapping frame
r^. (t) Is the real time GPS position vector in mapping
frame
R) (t) Is the attitude matrix between the body frame and
mapping frame
R^ Is the rotation matrix between the camera frame
and body frame
E" jc Is the GPS to Camera lever arm vector
Hy Is the Image point scale factor
ri Is the image point
t Is the time of exposure
For each image, rM
M (t) and RY(t) are obtained from the
c
P
determined from the: feature track extraction process in as
explained in Section 2.2. R^ and rj, are fixed values that
are obtained through calibration. r" and u^ are obtained
p
direct georeferencing system and any r? for hot spots are
through performing a 3D space intersection.
3. RESULTS
To test the overall accuracy of the F?D system, an airborne test
flight took place over controlled fire pits in Calgary in July 30,
2002. The controlled and monitored fires were set in fire pits at
the Bowness Park picnic area. The F?D system was mounted in
a remote sensing aircraft (aircraft and flight services provided
by Geodesy Remote Sensing, Calgary). Two multi pass test
flights at varying aircraft altitude took place over the test filed
to allow for collection of data under both day and night
conditions. Figure 8 shows a sample of the fires that were used
as targets, and the corresponding forest coverage of that target.
Any where that the blue sky is observed in the right hand image
is where the thermal imager could potentially see the fire.
Figure 7: Sample Target Hot Spot and Forest Canopy Coverage
Table 2 lists the positional accuracy of the aircraft derived
position from the WAAS and the OmniStar real-time systems
during the two test flights. The reference trajectory for these
results is the post processed double Differenced GPS (DGPS)
solution (accurate to 10 cm). The DGPS solution was derived
Intern
from th
(http://w
The tabl
is in the
Figure &
flights.
processe
KINGSE
attitude :
2 a
N e
e
Degrees Difference
e
Figure
Utilizing
Section
the FD
fire pits
processe
results
georefer
Pass 1
N
Pass
Ua
Pass
Pass 4
Un
Pass