Vous êtes sur la page 1sur 8

Stand-off detection of hidden threat objects on personnel at

checkpoints and in public areas using active millimetre-wave

Rory Doyle‡, Brendan Lyons‡, Alan Lettington*, Tony McEnroe‡, John Walshe‡,
John McNaboe‡, Peter Curtin‡, Stan Bleszynski+
‡ * +
Farran Technology Ltd The University of Reading S&M Process Control
Ballincollig Whiteknights Technology Inc.
Co. Cork Reading RG6 6AF, 1668 Lakehurst Rd,
Ireland UK Lakefield
Tel: + 353 21 4872814 K0L2H0, ON, Canada
Fax: + 353 21 4873892


The use of millimetre-wave imaging to identify threat objects such as guns and bomb material concealed on the person
is well documented. However, the technology has been hindered by the performance and cost barriers typically
associated with imaging at mm-wave frequencies. A novel scanning technique that minimises the receiver count while
operating at very high efficiency levels has made it possible to build a cost-effective and high-performance mm-wave
imager that can make security screening a commercial reality. The imager design allows for either passive or active
operation and its compact form factor is suitable for practical installation in security channel situations. The uses of this
technology include portal screening of personnel for high-resolution imaging of concealed threat objects or longer
distance surveillance type monitoring of checkpoints and crowds. This presentation details the use of the imager in an
active configuration to observe a checkpoint or crowd scene at stand-off distances of up to 50 metres. Target objects to
be detected are the hidden metal components associated with suicide bomb constructions. A typical bomb consists of
several explosive filled pipes strapped to the body or clusterings of small metallic objects embedded in explosives.
Trials at 94GHz have yielded positive results by showing the presence of concealed metallic objects on people at
distances of 25 metres. Objects detected have included simulated bomb constructions such as groups of metal pipes and
clusters of nuts and bolts. These tests have been conducted using a Gunn based CW source and direct detect receiver
unit. Further enhancement of the system includes the use of an FMCW front-end configuration.

Keywords: millimetre-wave, surveillance, stand-off, threat-object, radar, active, scanning


Imaging in the millimetre-wave region of the spectrum offers the unique property of being able to ‘see through’
obscuring materials such as clothing and cardboard. This property opens up application in the security area for scanning
personnel for concealed threat objects or contraband. In outdoor situations, the ability of millimetre waves to penetrate
fog, smoke and other atmospheric obscurants offers the potential for imaging of the surrounding terrain where visible
and often infrared imagers have limited or no visibility. To exploit the potential of millimetre-wave imaging, a suitable
receiving system must be employed to detect and process the appropriate signals in the 100GHz region of the spectrum.
To date, use of sensors for detecting signals at these frequencies has been hindered by cost and performance drawbacks
that have limited the effective use of millimetre-wave imaging. The development of monolithic microwave integrated
circuits (MMICs) operating up to 100GHz and scanning systems that gather scene information onto a sub-array of
receivers has opened up the possibility of commercially affordable and practically deployable millimetre-wave imaging
Imaging a scene using millimetre-wave frequencies can be can be done using a fully populated focal plane array.
However the cost and complexity of implementing a full array means only a limited number of applications can be
addressed with this approach. A more practical option is to use a single or sub-array of receivers and to scan the scene
using an electronic or mechanically driven approach. Using a mechanically driven optical scanning method allows for a
rapid capture of a given field of view in an efficient and easily interpreted user format.

The system described in this paper is a mechanically scanned millimetre-wave imaging camera that can be
interchangeably used in an active or passive mode and provides rapidly refreshed, raster imagery with high efficiency
across all millimetre wave frequencies. The system is focus adjustable to image close-in for portal screening application
or at long distances for use in surveillance situations. The emphasis in this paper is on long distance imagery for
detection of threat materials on personnel in stand-off surveillance use.


A mechanically scanned optical system for use in millimetre-wave imaging must address a number of key performance
targets. In order to achieve a suitable spatial resolution, an appropriate optical aperture must be used. The image refresh
rate must be made as rapid as possible through the use of a sufficiently fast line or frame scan while maintaining a
suitable integration time per pixel. The quality of the optical elements must be maintained to minimise loss and scatter
in the optical path. With these constraints in mind, the overall system needs to be compact so that it can be practically
deployed in a security or surveillance style application.

The system that is described here was specified for use in both portal screening and long distance surveillance
applications. The basic parameters are described in table 1. The system uses a rotating mirror configuration to
implement the scene scan with optical elements that are all purely reflective. By combining tilted plane and concave
mirrors whose rotations are equal and opposite a line scan of the scene is produced – one upward and one downward
scan for each complete rotation of the mirrors. A third mirror is used to pan the line scan across the scene. This panning
action builds up a raster frame of the scene.

Principle of Tilted rotating mirrors that focus a line-scan of the scene onto the millimetre-wave
operation receiver(s)
Optical elements Rotating mirrors – 100% reflective.
Optical aperture Concave focussing mirror – 600mm aperture
Field of view ±10° in the vertical (mechanically variable). 0 - 60° in the horizontal (programmable)
Angular resolution 0.3° at 94GH z
Refresh rate 1Hz for four channels – target of 10Hz for 16 channels
Dimensions L=1580mm x W=860mm x H=1620mm

Table 1: Main parameters of the millimetre-wave camera

The choice of a 600mm aperture was made based on the desired spatial resolution and the practical limit of rotating the
offset mirror at high speeds. This optical aperture fixes the achievable resolution and also determines the overall
physical size of the system.

The patented scanning approach1 has a number of key benefits – listed in table 2 – that result in a high quality image
being captured. A significant objective for millimetre-wave imaging to become a commercially realistic technology is to
achieve a price/performance threshold that will make it feasible for use in high throughput applications such as
checkpoints and airport security gates. The advances described in this camera show that millimetre-wave imaging is
capable of meeting the detection performance and cost objectives that will make it a technically and commercially
viable approach.
Feature Benefit
Reflective optics The scanning camera uses only reflective optics in directing and focusing the incident
radiation. This ensures that almost all of the signal reaches the receiver – maximising the
system efficiency and signal to noise ratio for the processing electronics.
The fact that there are no frequency dependent optical elements in the signal path means that
the camera is independent of the frequency being detected and can be used interchangeably
over the entire millimetre-wave spectrum.
High The line-scan is generated by rotating two circular mirrors at constant speed – each going in
rotation/refresh one direction only. This configuration allows for a high rate of revolution of the mirrors and
rates consequently a fast refresh rate in the image. Based on an array of 16 receivers it is planned to
produce images at a 10Hz refresh rate.
Raster scan The rotation of the mirrors generates a line-scan of the scene. Combining this with the
panning motion of a flapping mirror produces a raster frame of the scene. The uniformity of
the raster output enables image processing to be performed directly on the output.
Active and Because there is no manipulation of signal in the optical path the camera can be configured as
passive operation either a passive receive only unit or as an active transceiver system.
Low Cost The low receiver count and simple mechanics combine to provide a physically compact
millimetre-wave camera with high performance and low cost of manufacture.

Table 2: The principal benefits of the mechanical scanning system used in the millimetre-wave camera.

The operation of the system is illustrated in figure 1a where the raytrace shows the optical path from the target to the
receiver. Figure 1b shows a cutaway view of the mirror configuration and their relative motion while figure 1c shows
the finished system.


Plane mirror


Concave mirror

(a) (b) (c)

Fig 1: Operation and construction of the millimetre-wave imager. The optical path is shown in the raytrace diagram (a) while the
cutaway (b) shows the mirror, electronics and drive configurations within the frame of the camera. The finished assembly is
shown in the photograph (c).

The millimetre-wave scanner has so far been used in both active modes. In the case of passive imaging the target
application is personnel screening where the field of view is of portal dimensions and the distance is in the order of
1.5m from the camera. Imaging was carried out using 94GHz direct detect receivers and 140GHz heterodyne receivers.
Both options provided good results with expected trade-offs in resolution and sensitivity for the different frequency
options. Refresh rates for the system based on a 1m x 2m field of view were 4 seconds for a system with a single
channel receiver and 1Hz in the case of a four channel system.

Initial trials were conducted out of doors where the cold sky background was used to provide the thermal contrast for
successful imagery. Moving to an indoor scenario required that a thermal contrast be artificially generated. A number of
options are possible and methods such as noise sources, cold and hot loads were evaluated. Indoor imaging has been
successfully demonstrated as shown in examples in figure 2 where outdoor and indoor images are compared. In these
images concealed metallic materials simulating threat objects have been detected.

(a) (b)
Figure 2: Examples of indoor and outdoor millimetre-wave imagery. The picture (a) on the left
shows a concealed metal object on the person when outdoor (cold sky) contrast is used to
illuminate the subject. An example of indoor image (b) uses artificially generated illumination to
highlight metallic (blade, watch) and leather (wallet) objects.


In order to image at greater distances than those used for portal imaging, the scene needs a greater level of illumination
than can be provided through the contrast of cold sky or hot/cold load panels. Active sources of microwave energy are
needed to identify targets of interest at distances of up to 50m. The application areas for this type of imaging are the
identification of individuals in a crowd or at a checkpoint who are carrying potential threat objects that are concealed
inside clothing, baggage etc. A typical threat in this category is a large metallic weapon or a bomb consisting of metallic
pipes and shrapnel. In this situation the monitoring camera and personnel must remain at a safe distance during
observation so that they are sufficiently out of range of any detected weapon or explosive device.
The approach to imaging at stand-off surveillance distances is to actively illuminate the observed scene in order to
detect objects of interest. In order to achieve a sufficient level of resolution, the field of view must be reduced to
appropriate dimensions and is typically a few meters square at the distance of interest. Unless large apertures can be
used, diffraction limited spot size is generally large at the distances being observed.

The options for millimetre-wave imaging were to either use a discrete source that provided general scene illumination or
an integrated source such as an FMCW style configuration that uses the scanner optics to focus the illumination onto the
observed scene location. The two alternatives are illustrated in figure 3 below.

Passive FMCW
Imager Imager


Range: Typically 20 – 30m Range: Typically 20 – 50m

Source: 94GHz Gunn oscillator/Noise Source: 94GHz/77GHz FMCW radar

Figure 3: Active imager formats using separate and integrated sources of scene illumination

The sources used for the general illumination case were a coherent Gunn oscillator tuned to 94GHz centre frequency
and a wideband noise source. The noise source was useful in illuminating scene objects for distances of up to a few
metres. However, for the longer distances of interest it did not have sufficient power density to produce the required
return signal. Attempts to focus the output of the noise source onto the scene by injecting it through the system optics
were not successful most likely due to alignment difficulties in getting the origin of the source to be coincident with the
receiver position.

Using a 94GHz tuned Gunn oscillator, a general broadcast illumination of the scene was set up. The working distance
was set to be 20m from the imager. Targets used were simulated bomb components - aerosol canisters to simulate pipe
bomb constructions and clusters of nuts and bolts to simulate shrapnel. Control of the source power level was a critical
factor as the scene was easily saturated by relatively low emission levels. The final setting to achieve discrimination of
the target was a very low power – close to 0dBm. Images were taken of control conditions where a person not carrying

Location Marker

(a) (b) (c)

Figure 4: Example of active imaging for stand-off surveillance applications. The visual scene is shown in (a) with an
inset of the aerosol cans used to simulate pipe bomb materials. The scene imaged with passive radiometry shows no
evidence of the person or the concealed metallic objects. Once the Gunn oscillator is energised (c), the concealed metallic
objects are detected on the person as the bright return in the image.
Location reference markers

Figure 5: An example showing the capability of the camera to identify the individual on the right who is carrying
an assortment of metallic objects (nuts, bolts etc) concealed under clothing. The bright echo corresponds to the
person carrying the metal whereas the person on the left does not show up in the image. The dark areas in the
millimetre-wave image are the reflection of sky off of the metal strips that are used as location references.

metal stood in the scene and then of the same scenario except with the person concealing various metallic objects.
Figures 4 and 5 show example outputs of these trials where the ability to detect personnel carrying suspect items is

Trials were also conducted with FMCW radar front ends where the system optics is used in the transmission and
focussing of the outgoing illuminating signal. Transceiver units working at both 77GHz and 94GHz have been designed
for imaging in standoff surveillance applications. The use of FMCW radar introduces the possibility of gathering range
as well as intensity information from the scene. This additional range information opens up significant possibilities in
pinpointing people carrying threat objects particularly in crowd situations where two-dimensional intensity patterns may
not provide sufficient information to identify a threat location.

A number of system level optimisations had to be carried out in order to use the optical path for both transmit and
receive directions. In particular the elimination of internal reflections and the removal of the receive input from the
signal paths had to be accommodated in order to remove artificially induced defects from the images.

By using a focussed FMCW it was possible to get a good accuracy on locations of concealed threat objects in the scene.
Limitations on the VCO quality placed an outer boundary of approximately 20 metres on the working distance of the
radar. However, once this has been corrected it is expected that distances of over twice this range will be possible. The
field of view of the imager was restricted to give a view of approximately 4 metres square at the 20 metre distance.
Actual scanning time for such a window would be in the order of 2 – 4 seconds in an optimised system – redundancy in
the current configuration meant a large part of the sweep was not used for actual imaging.

The results confirmed the expected capability of the system to distinguish between people carrying metallic objects and
those without such material. An example of the radar output is shown in figure 6 where two individuals are screened and
the one carrying a number of concealed aerosol canisters is clearly identified. The corresponding range information was
also correlated to the location of the subjects.


The areas where long distance millimetre-wave imaging will find application centre on the identification of threat
objects at large stand-off distances. This early identification affords the user a level of safety once the detection has been
made. The typical situation would be the surveillance of a checkpoint or gateway where people are passing through a
Location marker

Figure 6: An example of the use of FMCW imaging to identify people carrying threat objects. The top left image shows the
normalised output from the radar where the dark background indicates areas of low level response. The lighter shadings show
areas of return where metallic objects may be present. The corresponding visual photograph is shown on the top right. The person
on the right hand side is concealing several aerosol canisters (similar to those in the figure 4 inset). Removing the low level returns
and overlaying the peak responses onto the visual picture gives an indication of a potential user interface (lower picture).

controlled area. In this case a continuous monitoring of the checkpoint can be set up by making the field of view
equivalent to that through which the people are passing. An alternative situation is where a more dispersed group of
people need to be monitored such as in an open public area or on a street. In order to monitor such a scene with a real
time refresh rate it may be more practical to sweep a narrow window field of view across the scene. Combining a swept
monitoring window with a visual video view of the scene would give a very effective user interface. This is illustrated in
figure 7. While both the broadcast illumination using the Gunn oscillator and the focused method using the FMCW have
been shown to work for these applications, the addition of range information provided by the radar gives it a key
advantage. In particular the public area application where identification of an individual in a randomly moving crowd
situation would need range data to accurately locate a person carrying a suspect item.


Figure 7: A representation of the use of a surveillance imager is shown here. The scanned window is swept across a
crowded public area. The millimetre-wave output is fused with its corresponding video image of the scene. By
showing only the high level returns from the scene as an overlay on the video display a very easily interpreted user
interface can be implemented.

Having demonstrated the ability to detect suspect materials carried on people at a distance, the evolution of the system
to a field deployable imager is an immediate objective. Depending on the intended usage, the imaging camera can be
configured to suit the applications described in the previous section. The fusing of millimetre-wave imager with a visual
picture of the scene has already been successfully demonstrated for still images. The extension of the camera to real
time imaging and the fusion of this output with a standard video feed will lead to a readily usable product.

Trials to date have worked on surveillance distances of up the 30 metres. The extension of this distance to 50 metres
will be demonstrated once upgrades to both the front-end and camera performance have been implemented. To date the
camera has been used for surveillance applications at both 77 and 94GHz. Further work with different frequency ranges
is expected to yield a better characterisation on the detection and discrimination of different target types.


The capability to identify concealed threat objects in stand-off surveillance situations has been demonstrated. Different
options for illumination of distant targets have been evaluated and both broadcast CW and focussed FMCW approaches
have been shown to be valid. The successful identification of concealed metallic targets at ranges of greater than 20
metres has been carried out and the extension to distances up to 50 metres is planned. User interfaces that will allow the
easy interpretation of millimetre-wave imagery have been proposed and implemented in a still image format.

The application of millimetre-wave imaging to stand-off security screening offers an effective means of monitoring
personnel at distances that can maintain safety for the observer and give adequate warning of potentially dangerous


1. A. H. Lettington, D. Dunn, N. E. Alexander, A. Wabby, B. N. Lyons, R. Doyle, J. Walshe, M. Attia, I. Blankson,

“Design and development of a high performance passive mm-wave imager for aeronautical applications” Proc. SPIE
Vol 5410, pp 210-218, Orlando, April 2004.