Aerospace and Electronic Systems Magazine September 2016 - 9

Wilson et al.

Figure 2.

The prototype SAA system was fitted as a payload to the nose of a
ScanEagle.

quired attitude accuracy level in order to minimise the residual jitter.

Morphological-Temporal Processing
Morphological processing techniques have been shown to be powerful tools for identifying small point-like features in images. It is
known that sophisticated morphological processing methods can be
formed by combining dilation and erosion morphological operations, where the size and shape of the image features to be highlighted can be controlled via the structuring element parameter [13].
The combination of morphological processing and temporal hidden Markov model (HMM) filtering techniques has been
demonstrated to achieve practical aircraft detection ranges at very
low false alarm rates [10]. Initially, the image data is preprocessed
with a bottom-hat morphological filter to detect point-like features.
Temporal filtering is then employed using a bank of four independent HMM filters. The success of the technique was demonstrated
onboard a general aviation aircraft [10], [14], [15]. Our challenge
was to prove the technique's real-time efficacy onboard the ScanEagle unmanned aircraft.

Detection Processing
After processing each image the detection algorithm provides a detection metric, which indicates the confidence that a threat aircraft
is present. Once the detection metric crosses a specified threshold,
a tentative detection is declared. If the detection metric is above
the threshold more than M times in the last N frames, a confirmed
detection is declared.
The detection threshold is a design parameter that affects the
sensitivity of the algorithm. A lower threshold will result in earlier detection but a higher false alarm rate. Conversely, a higher
threshold will result in a later (or missed) detection but a lower
false alarm rate.

THE SCANEAGLE UNMANNED AIRCRAFT
The Insitu ScanEagle is a small unmanned aircraft that is often
used to provide real-time imagery for a variety of applications.
SEPTEMBER 2016

Figure 3.

The Airborne Systems Laboratory (ASL) is a Cessna 172R. The ASL was
used as the threat aircraft for the SAA flight test and evaluation campaign.

As of late June 2015 the ScanEagle fleet had achieved more than
832,579 operational flight hours. Civilian applications of the ScanEagle have included bushfire surveillance, fisheries monitoring,
and marine mammal surveys [16], [17].
A number of the ScanEagle's key specifications are shown in
Table 1. The ScanEagle is launched using a pneumatic catapult.
A vertical rope suspended by a "SkyHook" is used to capture the
ScanEagle. Thus, the ScanEagle does not require a runway, which
means the ScanEagle can operate in a wide variety of locations.
Figure 2 shows the SAA system fitted as a payload to the nose
of the ScanEagle. The ScanEagle is shown hanging from the rope
provided by the Skyhook. One important aspect of the project was
the design and construction of a "nose" to mount the SAA system
onto the ScanEagle. The nose protected the SAA system from the
forces experienced by the aircraft during the catapult launch and
the subsequent recovery by the Skyhook.

THE AIRBORNE SYSTEMS LABORATORY
Figure 3 shows the Airborne Systems Laboratory (ASL), which
is a Cessna 172R [18]. The ASL is operated by the Australian Re-

Table 1.

Key Aircraft Specifications for the ScanEagle and
Cessna 172R
ScanEagle

Cessna
172R

Wing span (m)

3.11

11.0

Length (m)

1.55

8.28

Take-off weight
(kg)

1111

Cruising speed
(knots) [typical]

110

IEEE A&E SYSTEMS MAGAZINE

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine September 2016