Data Stream
Dual Frequency Air-Qualified
Antenna
DX4/75 Laptop Computer
Tilt Sensor Array
Ashtech Z12 Dual Frequency
GPS Receiver
Feedback Loop
PCMCIA A/D Conversion
Card With Timer Chip
weans ejeq
Photographic Operator
2
Zeiss UMK 10/1318 Universal
Camera and Control Box
Figure 1 - Overview of the Nottingham approach to GPS-Photogrammetry Integration.
but GPS would still be useful in reducing plan control
requirements (RMSE plan =+0.025m). Reduced highway
heighting precisions of between +0.010m and +0.020m
would obviously relax these requirements.
3. DEVELOPMENT OF AN INTEGRATED GPS-
CAMERA SYSTEM
There have been several stages to the development of
the system. The Stage One investigations of system
development (integration methodology and initial flight
trials) were fully reported to the United Kingdom
Photogrammetric Society (Smith and Joy, 1995a) and to
an ISPRS-FIG Joint Workshop in Barcelona (Smith and
Joy, 1995b). An integrated GPS Photogrammetry camera
system has now been developed from the original
prototype. Using the latest in receiver technology coupled
with in-house hardware and control software, perspective
centre coordinates can be derived at each instant of
exposure. Despite the use of an array of tilt sensors to
give approximate camera tilts, the main interest is in
position at this stage. As the motion of a helicopter is less
predictable than that of a fixed wing aircraft, interpolation
between GPS coordinate solutions would not be
desirable. Therefore, the Nottingham system uses the
Pulse Per Second (PPS) output of an Ashtech Z12
receiver to fire the camera on a GPS measurement
epoch. The delay between the rising edge of the PPS
output and the instant of exposure is compensated for
through the controlling circuitry without the need to modify
the UMK camera or control box. This delay was calibrated
in the laboratory with the typical camera settings and
orientation (to reproduce operational conditions) as
56.74ms with a standard deviation of 0.42ms. At the
usual airspeed of 15mph, this deviation corresponds to a
2.8mm error which illustrates the adequate repeatability
of the UMK 10/1318 mechanism.
Figure 1 provides an overview of the system showing the
dependancy of each component. Further information
concerning the refined system can be found in Smith et al
(1996).
The GPS position information is post-processed using the
‘On-The-Fly’ kinematic software developed independently
at Nottingham (Hansen and Joy, 1995). This software
uses a combination of two ambiguity search techniques
coupled with the option for direct resolution of the
widelane ambiguities. Such an approach significantly
reduces the number of integer combinations which must
be searched. Cycle slip detection and correction software
is also available within the IESSG for difficult portions of
data.
4. GROUND BASED EVALUATION TRIAL
An important part of the testing of the refined Nottingham
system was a ground based evaluation trial. By taking
photography of a large building facade whilst moving the
system on a trolley, truth positions could be calculated for
the antenna phase centre in a similar approach to that
reported in Hansen and Joy (1995) and widely
understood. A feedback loop in the system also enabled
recording of the pulse time, ensuring that the timer chip
was functioning correctly. Table 1 summarises the results
for two frames of photography which were observed twice
(observations a and b).
Frame | dX(m) | dY(m) | dZ(m) | dL (m)
4 4a 0.036 0.046 0.053 0.076
4 4b 0.039 0.041 0.050 0.075
3 1a 0.048 0.016 0.030 0.058
3 1b 0.037 0.012 0.031 0.049
816
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
Table 1 - Summary of Position Discrepancies.
It can be seen from preliminary evaluation of the vector
length (dL), that the relative accuracy is approximately
6cm. However, such a measure is affected by the
accuracy of the GPS processing which can easily vary at
the centimetre level. The photogrammetric accuracy is
