Sensors and Actuators A 174 (2012) 7584

Contents lists available at SciVerse ScienceDirect

Sensors and Actuators A: Physical
j our nal homepage: www. el sevi er . com/ l ocat e/ sna
Design of an electromagnetic imaging system for weapon detection based on
GMR sensor arrays
Gui Yun Tian, Abdalrahman Al-Qubaa

, John Wilson
School of Electrical, Electronic and Computer Engineering, Newcastle University, Newcastle upon Tyne, UK NE1 7RU
a r t i c l e i n f o
Article history:
Received 10 May 2011
Received in revised form
21 November 2011
Accepted 24 November 2011
Available online 4 December 2011
Keywords:
Electromagnetic devices
Magnetic sensors
GMR sensor
Electromagnetic imaging
Weapon detection and classication
a b s t r a c t
Concealed weapon detection is one of the most challenging issues facing the security community. It has
been shown that each weapon can have a unique ngerprint, which is a set of electromagnetic (EM)
signals determined by its size, shape, and physical composition. Extracting the signature of each weapon
is one of the major tasks in any detection system. This paper addresses the design of a detection system
for the identication of conductive objects based on their response to EM elds. The system consists of
commercial Walk-Through Metal Detector (WTMD) and a Giant Magneto-Resistive (GMR) sensor array
has been designed and built. Also, this paper describes how to construct a two-dimensional image from
the measured signals to be used for image processing purposes. The system validity is then checked based
on two concepts: data validation and multiple object separation. Finally, initial experimental work on the
automatic detection and classication of different metallic objects has been carried out. The promising
results indicate the feasibility of using this EM imaging method to identify objects.
Crown Copyright 2011 Published by Elsevier B.V. All rights reserved.
1. Introduction to imaging and weapon detection
technologies
In light of the abuse of guns and knives, automatic detection
and characterisation of weapons has attracted much attention in
recent years. Many approaches and systems have been proposed
and realised for security in airports, stations, law courts, etc. The
fact that most weapons are made of metallic materials makes
electromagnetic (EM) detection methods it the most prominent
methodology and systems built on the principle of EM induction
have been prevalent for many years for the detection of suspicious
metallic items carried covertly.
The Walk-Through Metal Detector (WTMD) is a commonly used
device for detecting metallic weapons and contraband items using
EM elds. Most WTMD units use active EM techniques to detect
and classify metal objects, see Fig. 1 [15]. An active EM eld,
in this instance, means that the detector sets up a eld using a
source coil, this eld is then used to probe the environment. The
applied, or primary, eld induces eddy currents in the metal under
inspection, which then generates a secondary magnetic eld that
can be sensed by a detector coil. The rate of decay and the spa-
tial behaviour of the secondary eld are determined by the targets
electrical conductivity, magnetic permeability, shape, andsize. Sets

Corresponding author.
E-mail address: abd.qubaa@ncl.ac.uk (A. Al-Qubaa).
of these measurements can be then taken and used to recover the
material, position, size and shape of the objects.
EM imaging has potential applications in medicine [6] and
industry [7] using the Magnetic Induction Tomography (MIT) tech-
nique. MIT is an imaging systembased on the same principle as the
WTMD. It is basedonanimages spatial distributionof the electrical
conductivity and magnetic permeability of objects [8]. MIT applies
a magnetic eld froman excitation coil to induce eddy currents in
the material to be studied; the magnetic eld from these is then
detected by sensing coils. MIT is sensitive to all three passive EM
properties: conductivity, permittivity and permeability.
Fig. 2 shows how a ferrite and copper bar can be distinguished
using MIT [4]. The image is reconstructed using an inverse algo-
rithm [9], however this solution is frequently poor because of
the nonlinear relationship between scattered eld and object. A
poor solution means poor image quality and consequently reduced
detectability [10].
This magnetic imaging technique has been used to detect and
discriminate between objects using several receivers and a sin-
gle transmitter as detailed in [11]. The spatial resolution of this
prototype system is presently 5cm. Consequently, the image of
a handgun would not be precise but it would be discernible. The
inverseprobleminthis systemis solvedbyacomplexmathematical
process, which adds more complexity.
Microwave imagers have been developed based on the EM
Reectometer (EMR) principal, EMR measures the EM waves
reected from an item in the wave-illumination region. Images
can be built up by scanning the device over the object under
0924-4247/$ see front matter. Crown Copyright 2011 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.sna.2011.11.034
76 G.Y. Tian et al. / Sensors and Actuators A 174 (2012) 7584
Fig. 1. (a) PMD2 WTMD metal detector [2]. (b) Diagramof a metal detector with an object inside the detection space [3].
Fig. 2. Discrimination between a ferrite bar and two copper bars using the MIT
system[4].
inspection, or using an array of antennae comprising of multiple
pairs of transmitters and receivers [12], as shown in Fig. 3. An
antenna array is moved around a person by a cylindrical shaped
mechanical scanner. The scan takes 47s before 3D cylindrical
holographic images are produced. The holographic imager can
detect threats such as weapons constructed of metal, plastic and
ceramic as well as explosive solids and liquids. Although these
systems can produce very impressive results, a constrained envi-
ronment is required.
Currently available weapons detectionsystems primarily detect
metal objects, they are either large in the case of WTMD, giv-
ing the possibility of circumvention, or for the hand-held metal
detector require close proximity to the person being searched,
putting the operator at risk [13]. Additionally, the WTMD works
with an adjustable threshold to discriminate between threat items
and personal items, depending on the mass of the object, which
can increase the false alarm level, see Fig. 4 [14]. Furthermore the
human body itself affects the sensitivity of the detector, so when
dealing with a material with a low conductivity or small size, the
human body may mask the signal of the object causing the material
to pass undetected, resulting in a poor reliability of detection [15].
There are several other EMtechniques used for gun detection such
as: millimetrewaves (MMW) [16], terahertz (THz) imaging[17] and
infrared imaging [18]. Each of these techniques has advantages and
disadvantages.
It can be concluded that the current EM imaging systems have
several drawbacks; lowimageresolution, theshapeof theEMsignal
may not correspondto the actual shape of objects, poor detectionin
a multiple object scenario, high cost, privacy invasion issue (when
usingMMW andTHz), harmful (usingX-ray) andthesignal received
only corresponds to the metallic part of the material which may
hinder the detection and classication of concealed weapons. All
these increase the false alarmrate of EMdetection.
This paper details the design and operation of a new metallic
object detection system utilising an array of Giant Magneto-
Resistive (GMR) sensors in conjunction with pulsed excitation to
develop a new WTMD for deployment in unconstrained environ-
ments, i.e. without users divesting themselves of metallic items.
This paper also describes how to construct a two-dimensional
image from the measured signals to be used for image processing
purposes. The system validity is then checked based on two con-
cepts: data validationandmultiple object separation. Finally, initial
experimental work on the automatic detection and classication of
different metallic objects has been carried out.
Fig. 3. A microwave imager for body inspection: (a) the inspection system, and (b) microwave images of a person carrying two concealed guns [12].
G.Y. Tian et al. / Sensors and Actuators A 174 (2012) 7584 77
Fig. 4. Cumulative signal effects in an active walkthrough weapon detector [14].
The rest of this paper is organised as follows: Section 2 will
describe the basic theory for EM imaging systems. Section 3 will
consider the systems design and principle of operation. EM eld
imaging is explained in Section 4. Systemvalidation and an exam-
ple of possible feature extraction and classication techniques are
presented in Section 5. The paper is concluded in Section 6 with
potential future work.
2. Theory for electromagnetic imaging systems
The EMresponse to a material can be obtained by solving:
E =
B
t
(1)
where E is the electric eld and B is the magnetic ux density. The
electric displacement and magnetic eld are introduced solely as a
matter of convenience when considering polarisable and magneti-
sable materials. The magnetic eld H is related to B through the
magnetisation M:
H =
1

0
B M(H) (2)
In this equation, M is explicitly written as a function of H. In
the material, the magnetisation vector Mis dened as the average
magnetic moment per unit volume. It is thus suitable to visualise
the magnetisation of a material as being froman assembly of mag-
netic dipoles. If these dipoles are distributed evenly throughout the
material, the material is consistently magnetised. For a nonmag-
netic material, suchas copper, there is nomagnetisation(M=0) and
thus, the magnetic ux density and the magnetic eld are related
by:
B =
r

0
H (3)
where
r
is the relative magnetic permeability of the target and
0
is the permeability of vacuum.
The functional relationship of the magnetisation with the mag-
netic eld, M(H), helps classify the three main classes of magnetic
materials: diamagnetic, paramagnetic, and ferromagnetic. The
magnetic eld at any point around the magnetic source is governed
by the following formula [19]:
B
x
(x, y, z) =

0
4
3
xz
(x
2
+y
2
+z
2
)
5/2
(4)
B
y
(x, y, z) =

0
4
3
yz
(x
2
+y
2
+z
2
)
5/2
(5)
B
z
(x, y, z) =

0
4
3
z
2
(1/(3(x
2
+y
2
+z
2
)))
(x
2
+y
2
+z
2
)
5/2
(6)
Fig. 5. Pulse induction metal detection.
78 G.Y. Tian et al. / Sensors and Actuators A 174 (2012) 7584
The magnetic eldproducedfromthe objects will be using same
equationbut withEqs. (4)(6) multipliedby the relative permeabil-
ity r.
The type of magnetic eld generated by its excitation coil is that
of a pulse induction eld. Pulse induction detectors typically pro-
duce a transmitter current, which is turned on for a time and then
turned off. The decaying eld generates pulsed eddy currents in
the target, which are then detected by analysing the decay of the
pulseinducedinthereceiver coil. Conductiveobjects showaunique
time-decay response. The pulse induction technique detects metal
objects by calculating the time-decay response of the pulse induced
in the receiver coil [20].
Fig. 5 illustrates the concept of magnetic inductive metal detec-
tionmethods usingthereceivedsignals for eachposition. Thegure
shows a change in decay rate of the signal received by pulse induc-
tion detector with respect to the reference signal when passing
over a metal object at position 10. The magnetic eld produced by
a source interacts with a nearby conductive object. The type and
strength of this interaction depends on several parameters such
as: the type of material that the object is made of, the size and
shape of the object, the orientation of the object in the magnetic
eld, the speed of the object through the magnetic eld, the dis-
tance between the sensors and the object. All these should be taken
into account when designing a system to detect and discriminate
between threat items [10].
3. Systemdesign and principles of operation
The system used for the experimental tests is based around an
array of NVE AAL002-02 GMR sensors [2123], used in conjunction
with the excitation coil in an ex-service CEIAWTMD. Table 1 shows
a list of the equipment used.
3.1. Magnetic sensor array specications
NVE GMR sensors were chosenfor the array. NVE offer a number
of sensors with a varying magnetic eld ranges. After initial tests
with Hall sensors (wide magnetic eld range but low sensitivity),
the AAL002-02lowhysteresis GMR sensor was chosenfor the array.
The sensors have a linear magnetic eld range of 1.510.5mT and
a sensitivity of 4.563V/T at a supply voltage of 15V.
The L in the sensor model name indicates that low hysteresis
(maximum 2%) GMR material has been used fabricate the sen-
sor, this was chosen because it was initially intended to utilise an
applied magnetic eld varying from0 to a maximumvalue where
the lower hysteresis value would minimise the error at low eld
strengths. However, after initial testing, it was found that a more
stable signal could be achieved by biasing the sensor response into
its linear region using a DC offset superimposed on the excitation
signal.
Fig. 6. Sensor spacing for array.
The spacing or separation of the array (Fig. 6) has a large impact
on the overall design of the system; the smaller the spacing, the
greater the number of sensors are needed and the greater the
complexity and cost of the system. Four different sensor spacings
were trialled during the tests: 7.5mm, 10mm, 15mmand 42mm.
After analysis of the results, the 15mm spacing was found to be
a good compromise between spatial resolution and system com-
plexity. The exaggeration of eld distribution for smaller objects
works to compensate for the sensor separation. Although the cho-
sen sensor separation means that the vertical accuracy can only be
guaranteed to be within 15mm, tests have shown that measure-
ment of the actual position of the distribution is not particularly
useful in object characterisation and analysis of other aspects of
the EMsignature are more reliable for object discrimination.
3.2. Systemconguration
Ablock and connection diagramof the systemis shown in Fig. 7.
The upper part of the diagramis duplicated ve times to make 80
channels (16 channels on each of the ve cards). Two 8-channel
sensor boards are connected to each of the 16-channel ampli-
er boards via a 20-core ribbon cable. The INA instrumentation
ampliers provide differential termination and amplication for
the sensor outputs. The amplier circuits are powered by a 15V
power supply. The outputs fromthe amplier boards are connected
to the data acquisition boards in the PC via the breakout boxes. An
additional connection is established to the data acquisition board
fromthe function generator. This allows the data acquisition to be
synchronised to excitation waveform.
A function generator supplies the excitation waveform. The
Bipolar power amplier is set to produce an output current that
is proportional to the input voltage supplied by the function gen-
erator. The output fromthe function generator must be connected
to the current programming input on the amplier to achieve this.
The output fromthe power amplier is connected to the coil in the
detector board via the arch control box. None of the electronics in
Table 1
Equipment list.
Equipment name Description
Agilent 33250A function generator Provides excitation waveformto power amplier.
Kepco BOP 36-12ML bipolar power amplier Provides excitation to the coil where the excitation current is proportional to the excitation voltage
from the function generator.
National instruments data acquisition system PC equipped with a PXI bus to accommodate multiple data acquisition cards.
5 NI PXI-6251, 16 input data acquisition cards. Allows acquisition of 80 channels of data at a sample
rate of 125kHz.
5 breakout boxes and cables to allowus to establish a connection to the data acquisition cards.
Sensor boards Each board contains 8NVE AAL002-02 GMR sensors.
Amplier boards Each board contains 16 circuits based on the INA111 instrumentation amplier, to allowconnections
fromtwo 8-channel sensor boards.
CEIA Walk-Through Metal Detector and control box We provide our own pulsed excitation to the coils in the metal detector panel through a connection in
the control box.
G.Y. Tian et al. / Sensors and Actuators A 174 (2012) 7584 79
Fig. 7. Systemblock and connection diagram.
the control box are used in the test; it is just there to establish a
connection to the detector panel.
Fig. 8a shows the positioning of the sensor array. The array is
aligned with the coil to pick up any distortions in the applied eld
due to the presence of metallic materials. Fig. 8b shows the inter-
action between the applied eld and any sensor in the array and
Fig. 8c shows the pulse response from a group of sensors. If no
object is present in the WTMD, the eld measured by the sensor is
unchanged; the presence of a metallic object causes a distortion of
the eld, which can then be measured by the sensor. Fig. 9 shows
the overall systemset up in the Laboratory.
3.3. Excitation eld response
In the system, pulsed excitation is applied to the coil. Pulsed
excitation provides the opportunity to apply an interrogating eld
Fig. 9. Systemsetup in the lab.
with rich frequency components in a single waveform. In the tests
detailed inthis paper, a pulse repetitionfrequency of 500Hz is used
with a pulse width of 1ms and an applied current of 0.51.5A.
Fig. 10 shows the pulse response for a steel object and an alu-
miniumobject measured using a single GMR sensor. It can be seen
from the plots that the change in pulse response for the presence
of an object (steel or aluminium) is actually very small. Comput-
ing the difference between the signal with and without an object
present, as shown in Fig. 10b and d (amplication 200), allows
us to accentuate the difference between the two signals. It can be
seen that a peak in the difference signal can be observed during the
rising/transient part of the signal; the time and amplitude charac-
teristics of this signal can be used to extract information about the
object under inspection.
4. Electromagnetic eld imaging
Different metallic objects were used for the conduction the
experiment andobservationof thesystems reactiontothreat items
(i.e. guns and knives) and non-threat items (i.e. mobile phone, keys,
Fig. 8. (a) Sensor array positioning with respect to coil, (b) interaction of applied eld and GMR sensor, and (c) uniformpulse response froma group of sensors.
80 G.Y. Tian et al. / Sensors and Actuators A 174 (2012) 7584
Fig. 10. (a) Pulse response for presence of aluminiumobject, (b) rising edge of pulse response for aluminiumobject with difference calculated, (c) pulse response for presence
of a steel object, and (d) rising edge of pulse response for steel object with difference calculated.
etc.). The resultant EMsignals measuredby GMR sensors during the
presenceof object inthesystemareorganisedas atwo-dimensional
array to be used for image processing purposes. In this section two
types of images construction are described: Max-value images and
Transient response images.
4.1. Max-value image
In this type of image construction the maximum value of each
signal measured by GMR sensors during the presence of object in
the systemis captured and recorded as follows:
1. The object under inspection is moved through the detector with
data being acquired at a pulse repetition rate of 500Hz (Fig. 11a).
2. Sets of 10 pulses are averaged to produce a single pulse response
(Fig. 11b).
3. A single value is computed fromeach pulse response (Fig. 11b).
Inthis case, the maximumvalue of the difference signal has been
used.
4. Each of these single values corresponds to a single pixel in the
nal image (Fig. 11c).
Thus, over time the EM eld distribution, as the object moves
through the array, can be determined. Fig. 12 shows some samples
used in the test and their constructed max-value images.
4.2. Transient response images
In order to extract more information about the objects in the
WTMD fromthe test results, a formof transient analysis has been
employed. In this transient EM signature imaging technique, the
pulse response from each sensor is analysed and chopped into
sections, or time slots, as shown in Fig. 13a. The values of the sam-
ples in each time slot are averaged, and using the data from all
sensors for the whole test, nally an image is built up for each
time slot.
Fig. 13bshows asequenceof thesetransient images for thehunt-
ing knife. Analysis of the transient image sequence can be used to
extract more information about the object under examination.
5. Systemvalidation
In this section the proposed system is validated by consider
three characters: data validation using repeatability, efciency of
systemmeasurement for multiple object separation, andsuitability
for classication techniques.
5.1. Data validation
Simple repeatability tests for the samples were carried out
where eachsample is testedseveral times tocheckdata validityand
the results are then plotted for all objects. Fig. 14 shows the ampli-
tude change of GMR signals for ve repetitions (Rep1toRep5) of the
test for six samples representing different real handguns borrowed
from the Metropolitan Police Force. It can be seen from the plot
that, as would be expected, the test has the greatest repeatability.
5.2. Multiple object separation
Toassess thecapabilities of thesystemtodetect multipleobjects
and determine the optimal object separation for object discrimina-
tion, the test shown in Fig. 15a was performed, with a replica gun
anda phone. Inthis test, the sample holder is employedto move the
G.Y. Tian et al. / Sensors and Actuators A 174 (2012) 7584 81
Fig. 11. EMimage constructed fromdata acquired fromline array over time.
objects through the WTMD in a controlled manner. The replica gun
is clamped in the sample holder and a mobile phone is hung next
to it at various separation distances. Fig. 15b shows the Max-Value
Image for the gun and phone at 120mmseparation.
It canbe seenfromthe toprowinTable 2that at a separationdis-
tance of 40mmbetweentwo objects are virtually indistinguishable
fromone larger object. Only at a separation distances greater than
60mmstart to be able to distinguish the two objects. This discrim-
ination of object signatures can be further enhanced by employing
a thresholding technique as shown in row2 of Table 2, where clear
object discrimination can be achieved.
5.3. Object classications
Classication techniques of grouping Principal Component
Analysis (PCA) features have been used to the system for the
classication of metallic objects in conjunction with previous
study using cross correlation of transient features [24].
Utilising the pulse response fromthe material under inspection,
the objects can be detected and the material can be classied into;
ferrous, non-ferrous or a combination of the two. Consequently,
through analysis of the image sequence generated by the transient
analysis, any object detected by the system can be classied. An
example of this is shown in our previous work on cross correlation
of transient features [24], where a cross correlation technique has
been applied to the transient image sequence where the maximum
cross correlation values are plotted after being sorted by ascending
amplitude and processed to classify the objects into param-
agnetic (aluminium), ferromagnetic (steel) and combinations
of both.
To investigate the system ability to discriminate between dif-
ferent objects of the same type and evaluate features, six real
Fig. 12. (a) Samples in the holder. (b) The equivalent EMimages.
82 G.Y. Tian et al. / Sensors and Actuators A 174 (2012) 7584
Fig. 13. (a) Pulse response with time slots marked, and (b) result of imaging the transient response fromthe hunting knife sample.
Fig. 14. Amplitude difference for ve repetitions of the test for the real gunsamples.
handguns were tested and different features have been investi-
gated in our study in [25]. PCA has been used as one of the feature
extraction technique as described in [25], preserving the total
variance of the original images. The rst two components, PCA1
and PCA2 have been plotted in 2D space in Fig. 16. The test
was repeated ve times with the gun in the holder being moved
through the WTMD and the results are shown in Fig. 16a. Like-
wise, the results for ve tests of a person walking through the
system with a handgun concealed inside his jacket pocket shown
in Fig. 16b. Results showthe efciency of systemmeasurements for
classication.
In comparison with conventional induction based WTMDs, the
GMR array based systemhas shown great potential in material dis-
crimination, as samples are made frommixed materials are clearly
distinguished. Whereas with currently induction based WTMD,
only discrimination between metal and non-metal is possible, our
novel systemhas taken previous possibilities a step further.
Fig. 15. (a) Multiple object test set-up, and (b) Max-Value Image for gun and phone for a separation distance of 120mm.
G.Y. Tian et al. / Sensors and Actuators A 174 (2012) 7584 83
Table 2
Thresholding technique applied to discriminate between the two objects.
40mmseparation 60mmseparation 80mmseparation 120mmseparation
Fig. 16. Discrimination using PCA for six guns repeated ve times: (a) gun in the holder and (b) gun inside a jacket pocket.
6. Conclusions and future work
An EM weapon detection system based on the application of
pulsed excitation in conjunction with accurate, high spatial reso-
lution magnetic eld sensing using GMR sensor arrays is designed.
Circuit design, data acquisition and real-time imaging of EMelds
have also been investigated. Furthermore, feature extraction and
image processing were conducted to check the systems capability
to discriminate between different objects. Real handguns and com-
monly used objects have been used as samples to test the system
andas a proof of concept. Inconclusion; this novel GMR array based
system has a greater scope for object detection and classication
than other current WTMD technologies.
A stand-alone walk-through system with superior object
discrimination and localisation capabilities is envisaged for future
exploitation. The discrimination capabilities of the system could
be developed to the point that individuals could pass through
the system without the need to remove metallic objects from
their person. This would be realised through training the system
to identify threat objects by presenting the system with a wide
variety of threat and non-threat objects and programming the
response accordingly. New feature extraction and classication
methods, such as neural networks will also be investigated.
Acknowledgments
This project is funded under the Innovative Research Call in
Explosives and Weapons Detection (2007), a cross-government
programme sponsored by a number of government departments
and agencies under the CONTEST strategy. The authors would like
to thank cross-government departments for the joint experimental
tests.
References
[1] A. Agurto, Y. Li, G.Y. Tian, N. Bowring, S. Lockwood, A review of concealed
weapon detection and research in perspective, in: Proceedings of the 2007
IEEE International Conference on Networking, Sensing and Control, London,
2007.
[2] Ofcial site of the CEIA Walk-Through metal detector 2011. http://www.ceia.
net/security.
84 G.Y. Tian et al. / Sensors and Actuators A 174 (2012) 7584
[3] N.G. Paulter, Users Guide for Hand-Held and Walk-Through Metal Detectors,
NIJ Guide 600-00, NCJ 184433, Ofce of Science and Technology, U.S. Depart-
ment of Justice, Washington, DC 20531, 2001.
[4] K. Kotter Dale, G. Roybal Lyle, E. Polk Robert, Detection and classication of
concealedweapons using a magnetometer-basedportal, in: Proc. SPIE, Sensors,
and Command, Control, Communications, and Intelligence (C3I) Technologies
for Homeland Defense and LawEnforcement, vol. 4708, 2002, pp. 145155.
[5] C.V. Nelson, C.B. Cooperman, W. Schneider, D.S. Wenstrand, D.G. Smith, Wide
bandwidthtime-domainelectromagnetic sensor for metal target classication,
IEEE Transactions on Geoscience and Remote Sensing 39 (2001) 11291138.
[6] R. Merwa, K. Hollaus, O. Biro, H. Scharfetter, Detection of brain oedema using
magnetic induction tomography: a feasibility study of the likely sensitivity and
detectability, Physiological Measurement 25 (2004) 347354.
[7] T. York, H. McCann, K.B. Ozanyan, Agile sensing systems for tomography, IEEE
Sensors Journal 11 (2011) 30863105.
[8] H. Grifths, Magnetic induction tomography, Measurement Science and Tech-
nology 12 (2001) 11261131.
[9] M. Soleimani, Computational aspects of low frequency electrical and elec-
tromagnetic tomography: a review study, International Journal of Numerical
Analysis And Modelling 5 (2008) 407440.
[10] N.G. Paulter, Guide to the Technologies of Concealed Weapon and Contraband
Imaging and Detection, NIJ Guide 602-00, NCJ 184432, Ofce of Science and
Technology, U.S. Department of Justice, Washington, DC 20531, 2001.
[11] B. Zollars, B. Sallee, M. Durrett, C. Cruce, W. Hallidy, Concealed weapons detec-
tion using low-frequency magnetic imaging, in: Proceedings of the SPIE, The
International Society for Optical Engineering, Conference on Surveillance and
Assessment Technologies for LawEnforcement, vol. 2935, 1997, pp. 108119.
[12] M.S. David, L.M. Douglas, E.H. Thomas, Three-dimensional millimeter-wave
imaging for concealedweapondetection, IEEE Transactions onMicrowave The-
ory and Techniques 49 (2001) 15811592.
[13] US Department of Justice, Final Report Demonstration of a Concealed
Weapons Detection System Using Electromagnetic Resonances, January 2001
[Online]. http://www.ncjrs.org/pdflesl/nij/grants/190134.pdf.
[14] J. Ashcroft, J.D. Deborah, V.H. Sarah, Walk-Through Metal Detectors for Use
in Concealed Weapon and Contraband Detection, U.S. Department of Justice,
Ofce Justice Programs, National Institute of Justice, 2003.
[15] H. Chen, S. Lee, M. Raghuveer, M. Slamani, K. Pramod, Imaging for concealed
weapon detection, IEEE Signal Processing Magazine 22 (2005) 5261.
[16] S. Yeom, D. Lee, J. Son, M. Jung, Y. Jang, S. Jung, S. Lee, Real-time outdoor
concealed-object detection with passive millimetre wave imaging, Optics
Express 19 (2011) 25302536.
[17] K.B. Cooper, R.J. Dengler, N. Llombart, B. Thomas, G. Chattopadhyay, P.H. Siegel,
THz imaging radar for standoff personnel screening, IEEE Transactions on Ter-
ahertz Science and Technology 1 (2011) 169182.
[18] S.-Y. Cho, N.-P. Tin, Using infrared imaging technology for concealed weapons
detection and visualization, in: TENCON 2010 2010 IEEE Region 10 Confer-
ence, November 2124, 2010, pp. 228233.
[19] G. Lehner, Electromagnetic Field Theory for Engineers and Physicists, Springer-
Verlag, Berlin/Heidelberg, 2010.
[20] C.V. Nelson, Metal detection and classication technologies, Johns Hopkins APL
Technical Digest 24 (2004) 6266.
[21] NVE Corporation, GMR sensor catalogue 2011. http://www.nve.com/
Downloads/catalog.pdf.
[22] J. Lenz, A.S. Edelstein, Magnetic sensors and their applications, IEEE Sensors
Journal 6 (2006) 631649.
[23] J.P. Sebasta, J.A. Lluch, J.R.L. Vizcano, Signal conditioning for GMR magnetic
sensors: applied to trafc speed monitoring GMR sensors, Sensors and Actua-
tors A 137 (2007) 230235.
[24] A.R. Al-Qubaa, G.Y. Tian, J. Wilson, Electromagnetic imaging systemfor weapon
detection and classication, in: Fifth International Conference on Sensor Tech-
nologies and Applications, France, 2011, pp. 317321.
[25] A.R. Al-Qubaa, G.Y. Tian, J. Wilson, Feature extraction, selection and fusion
of metallic object classication based on electromagnetic eld imaging, IEEE
Transaction Image Processing, submitted for publication, Paper ID: TIP-07951-
2011.
Biographies
Gui Yun Tian received the B.Sc. degree in metrology and instrumentation and M.Sc.
degree in precision engineering fromthe University of Sichuan, Chengdu, China, in
1985 and 1988, respectively, and the Ph.D. degree from the University of Derby,
Derby, U.K., in 1998. He then became a Research Fellow and Senior Research Fel-
low with the University of Derby and the University of East Anglia, U.K. From2000
to 2006, he was a head of the group of Systems Engineering, with the Univer-
sity of Hudderseld, U.K. Since 2007, he has been based at Newcastle University,
Newcastle upon Tyne, U.K., where he has been Chair Professor in Sensor Technolo-
gies and M.Sc. Program Director of Advanced Sensor Technology. Currently, he is
Group Head of Communications and Signal Processing in the School of Electrical,
Electronic and Computer Engineering, Newcastle University. He has published over
150 books and papers in English and Chinese in the above area. He has coordi-
natedseveral researchprojects fromthe Engineering andPhysical Sciences Research
Council (EPSRC), Royal Academy of Engineering and FP7, on top of this he also has
good collaboration with leading industrial companies such as Airbus, Rolls Royce,
BP, nPower and TWI among others. Dr. Tian is a Fellow of the Institution of Elec-
trical Engineers, U.K., and InstNDT. He was the recipient of the John Grimwade
Award from the British Institute of Non-Destructive Testing. He is also a Chinese
Changjiang scholar and as well as being on several editorial boards of international
journals.
Abdalrahman Al-Qubaa received the B.Sc. degree in mathematics and computer,
and M.Sc. degree in remote sensing centre/computer science from the University
of Mosul. Iraq, in 1994 and 2002, respectively, he work in ministry of Trade, min-
istry of Oil and ministry of Financial as a head of computer unite then he work as
a lecturer in the Mosul University tells 2008 and now he is a Ph.D. student in the
department of electrical, electronic and computer engineering in Newcastle Uni-
versity/UK. His main research interests are in electromagnetic signal and image
processing, electromagnetic in Non-destructive testing and weapon detection and
classication.
John Wilson received his Ph.D. fromdepartment of electrical, electronic and com-
puter engineering, Newcastle University in 2009 after completing a BEng (Hons)
in electronics and communication systems at Leeds Metropolitan University in
2004. Current research interests include Barkhausen acoustic emission and pas-
sive magnetic eld sensing techniques for non-destructive testing and evaluation
of ferromagnetic materials.