4-1-2 
1 INTRODUCTION 
Bushfires represent a significant natural disaster threat throughout 
much of Australia, and thus bushfire mitigation is a high priority 
for power utilities, both in regard to minimising fire damage to 
transmission line infrastructure and ensuring that the threat of the 
fire from vegetation overhanging high voltage transmission lines 
is minimized. The latter is achieved partially through ensuring 
adequate clearance between powerlines and adjacent vegetation, 
especially fire prone eucalyptus trees. 
Given the 1000s of kilometers of power lines crisscrossing the 
Australian landscape, the ongoing task of monitoring vegetation 
clearance is very significant. Indeed, in the context of asset 
management, not only clearance dimensions need to be recorded, 
but also the spatial locations of the power poles and the 
associated attributes (e.g. information regarding the insulators, 
presence of transformers, number of cross-arms, etc.). The 
recording of all these features can be achieved using the 
technology of mobile mapping. 
The Rapid Centreline and Attribute Mapping System (RCAMS) 
is a vehicle, railcar or aircraft borne rapid mapping system 
developed to enable the recording of positional and attribute data, 
using the integrated technologies of GPS satellite positioning, 
inertial positioning sensors and stereo video imagery (Leahy & 
Judd, 1998). The principal applications of RCAMS have to date 
been road centreline mapping (including the recording of 
roadside attributes such as signage) and powerline mapping. 
RCAMS has been successfully employed to provide in-car 
navigation data for major cities in Australia, and it has been used 
to record more than 2000km of high-voltage powerlines in the 
State of Victoria. 
As is indicated in Figure 1, the vehicle mounted configuration of 
RCAMS, while utilizing multi-camera video imagery, has not 
thus far employed stereo imagery since road attribute data is 
referenced only by the adjacent position on the road centreline. 
Stereo imagery has, however, been employed as a component of 
an RCAMS configuration utilised for high-precision, rapid rail 
mapping where rail corridor features are recorded to an accuracy 
of 0.3m (Hunt et al, 1998). With airborne RCAMS, stereo 
imagery plays a central role in feature positioning and in the 
present paper the discussion will be confined to this 
configuration. 
Put simply, in the application of airborne RCAMS to powerline 
and vegetation mapping, the position of any ground feature can 
be determined by spatial intersection from the stereo-video 
imagery, with the geo-referencing of this position being provided 
by the Exterior Orientation (EO) of the dual-camera system. The 
positional component of the EO is provided by GPS, whereas the 
orientation component is determined using both multi-antenna 
GPS and inertial navigation sensors. 
In order for both power pole positions and vegetation clearance to 
be determined to the required accuracy, nominally the lm level, it 
is important that a comprehensive sensor calibration and EO 
determination be performed for the dual-camera video system. 
This photogrammetric aspect of RCAMS forms the subject of the 
present paper. Following a brief overview of RCAMS, and a 
description of the video-based positioning component, the 
approach adopted for camera system self-calibration and EO is 
described. The results of an experimental field calibration are also 
presented and conclusions are drawn regarding the field 
calibration procedure and the merits of employing a mobile 
mapping approach for powerline vegetation mapping to support 
bushfire mitigation. 
2 BRIEF OVERVIEW OF RCAMS 
Figure 2 illustrates a schematic of the airborne RCAMS 
configuration. Aircraft position is recorded via an Ashtech 3DF 
GPS system, which incorporates four GPS antennas, a ‘reference’ 
antenna over the centre of the fuselage, an antenna on each wing 
tip and one on the tail. Orientation (roll, pitch and yaw) of the 
aerial platform is provided by the 3DF system, with pitch and roll 
rate gyros providing supplementary orientation data. Kalman 
filtering provides the fundamental computational model for 
integrating the positional and orientation data, as described in 
detail by Leahy and Judd (1998; 1999). Not shown in the 
schematic of Figure 2, but necessary for the kinematic positioning 
of RCAMS, is a GPS reference station in the vicinity of the 
project area. 
RCAMS incorporates video recorders for three video imaging 
cameras, a wing tip mounted, forward-looking stereo 
configuration, the calibration of which forms the main topic of 
this paper, and a camera aimed vertically downwards. In the 
powerline mapping project discussed, experimental application of 
a scanning infra-red CCD imager was also attempted. This met 
with limited success, however, due to the characteristics of the 
particular IR imager used. As a consequence, the vertical 
imaging capability of RCAMS will not be further discussed in the 
present paper. 
The two stereo cameras comprised Pulnix CCD sensors (pixel 
array size of 750 x 580 and format of 6.4 mm x 4.8 mm) with 
lenses of 17 mm focal length. Mounted with a photogrammetric 
base of 10.4 m, the cameras were mildly convergent so as to 
produce nominally 100% overlapping images when tilted 
downwards at a depression angle of 35° from the horizontal, and 
when flown at a low-level ground clearance of 50m. The 
resulting average camera-to-object distance of 80m yielded an 
image scale of 1:4700. As will be mentioned, digital stereo image 
pairs where extracted from the SVHS video tape at about 20m 
intervals, which provided an along-track image overlap of 60%. 
3 CAMERA SELF-CALIBRATION AND EO 
3.1 Network Configuration 
For practical reasons it was decided to adopt a system calibration 
approach in the determination of photogrammetric parameters for 
the airborne RCAMS. In order to geo-reference features 
identified in the stereo imagery, the following are required: