A Portable, Low-cost System for Optical Explosive Detection based
on a CMOS Camera
Ross N. Gillanders
1
, Iain A. Campbell
1
, Fei Chen
2
, Paulina O. Morawska
1
, Ifor D. W. Samuel
1
and Graham A. Turnbull
1
1
Organic Semiconductor Centre, SUPA, School of Physics & Astronomy, University of St Andrews,
Fife KY16 9SS, Scotland
2
Key Laboratory for Flexible Electronics, Institute of Advanced Materials (IAM), Nanjing Tech University,
30 South PuZhu Road, NanJing, 211816, P. R. China
Keywords: Explosives, Conjugated Polymer, Organic Semiconductors, Raspberry Pi, Imaging.
Abstract: Humanitarian demining requires a variety of methods and instrumentation for effective mine clearance,
since a wide range of materials are used in mine manufacturing. However, landmines release vapours over
time that can be detected, for example, by sniffer dogs. Optical sensor systems are especially suited to this
application due to the potential for lightweight, portable, low-cost systems that nevertheless have fast
response times and ppb-level sensitivity to explosive vapours. In this paper we present a system for
detection based on a low-cost Raspberry Pi platform with an integrated CMOS camera. The conjugated
polymers Super Yellow and Polyfluorene are excited by an LED, and the quenching effect by DNB vapour
is monitored by the camera to indicate the presence of explosives. The system shows potential as a user-
friendly, lightweight platform for explosive vapour sensing.
1 INTRODUCTION
The tools required for humanitarian demining must
be able to operate across a wide range of
environments, since legacy landmines remain in
countries with extremely disparate climates. These
climates can range from Mediterranean, such as that
in Croatia; to dry desert climates in Africa; to
monsoon-prone countries, for instance those in
South East Asia. Different tools for demining have
different degrees of success in the various
environments. The most common methods for mine
detection include metal detectors and sniffer dogs;
the combination of both is often used in the field.
However, these methods can have disadvantages:
metal detectors can identify harmless fragments of
metal, or, with potentially fatal consequences, be
unable to detect a mine made from another material
such as plastic. Canines have strict working
practices with the time the dogs can spend sniffing
for mines restricted by both guidelines and the
willingness of the dog at any particular moment
(Porritt et al., 2015). Other common methods are
inherently dangerous, such as prodders; thus there is
a need in humanitarian demining for advanced,
sensitive, robust instrumentation (Newnham and
Daniels, 2001). In this paper we present a portable
optoelectronic sensor based on explosive vapour
detection.
Mines can release small amounts of vapour for a
long time after being buried – for example,
trinitrotoluene (TNT) degrades over time to produce
dinitrotoluene (DNT), which then is released at
ground level. There has been extensive work done
on explosive sensing with conjugated polymers
(Wang et al., 2011a; Wang et al., 2011b; Wang et
al., 2012; Narayanan et al., 2008; Thomas et al.,
2007), which are electron-rich materials used in
organic semiconductor research. If nitroaromatic
explosive vapours adsorb onto the polymer surface
while these polymers are photoexcited by incident
light, electrons are transferred to the deficient
nitroaromatic vapour, resulting in loss of fluorescent
emission, as shown in Figure 1. The luminescence
can be monitored in real-time to provide a
quenching profile, thus indicating the presence of
explosive vapours.
One of the chief advantages to optical sensing is
that it can be achieved through relatively low-cost,
portable and fast-responding instrumentation. These
132
Gillanders, R., Campbell, I., Chen, F., Morawska, P., Samuel, I. and Turnbull, G.
A Portable, Low-cost System for Optical Explosive Detection based on a CMOS Camera.
DOI: 10.5220/0005739801300134
In Proceedings of the 4th International Conference on Photonics, Optics and Laser Technology (PHOTOPTICS 2016), pages 132-136
ISBN: 978-989-758-174-8
Copyright
c
2016 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
factors are crucial in humanitarian demining, since
the work is often conducted in challenging
environments. Additionally, the operators of such
devices will not necessarily be trained scientists or
technicians, so ultimately the operation of the
instrument should be user-friendly, with clear and
distinctive data visualisation.
Figure 1: Quenching mechanism for conjugated polymer
by explosive vapour.
While many fluorescence-based sensing
platforms successfully utilise low-cost electronics,
such as photodiodes, recent years have seen the
introduction of off-the-shelf consumer computing
platforms including the Arduino family of
microprocessors and the Raspberry Pi. Use of these
platforms in analytical instrumentation has gradually
begun to appear in literature (Bitella et al., 2014;
Tonacci et al., 2015; Leeuw et al., 2013), since they
offer advantages in cost, open-source hardware and
software, user-friendliness, small footprint, and
flexibility in deployment. Smartphones have
recently become tools for chemical sensors (Comina
et al., 2015), and since the CMOS camera used in
conjunction with the Raspberry Pi processor is
similar to those found in typical smartphones, the
technology presented here may eventually be
integrated with them, broadening and making more
accessible the system for communities in need of
easy-to-use, low-cost, portable optical sensor
systems for mine detection. While the change in
luminescence intensity from the sensor films can be
monitored using a standard silicon photodiode, this
generally allows only the measurement of a single
wavelength from a single film. The ability to
measure multiple luminescent sensor spots with a
single detector would present advantages in both
simultaneous sensing of multiple analytes, and in
selectivity to identify the, in this case, various
nitroaromatic groups. The use of a camera system
allows for this multi-analyte detection in the future
since the principle of the image processing, reported
here, can be applied to more than one emission spot
in a single image. This system could prove useful in
a “demining toolbox” in conjunction with other
technologies to allow a deminer to make a choice of
complementary systems appropriate for the
environment. For instance, a metal detector could
identify suspect sites, and the air sampled above
those sites for trace explosive vapours. Since the
system would also be suitable for sensing other
analytes, it may be useful for environmental, food
quality or other applications in the future.
In this paper we present an optical sensor system
using the Raspberry Pi platform in conjunction with
a CMOS camera, excitation LED, Polyfluorene
(PFO) and Super Yellow (SY) conjugated polymers
to detect dinitrobenzene (DNB) vapour. Early results
indicate that the system could ultimately be applied
for the detection of buried landmines.
2 EXPERIMENTAL
2.1 Film Fabrication &
Characterisation
Films based on Merck Super Yellow and PFO were
prepared by spin-coating a 6.5mg/ml solution of the
polymer in toluene at 2000rpm onto a glass slide.
The film thicknesses were measured with a Veeco
Dektak 150 surface profiler and found to be 100nm
thick on average. Absorption and
photoluminescence spectra were measured,
respectively, using a Cary 300 Bio UV-Vis
spectrometer and an Edinburgh Instruments FLS980
Fluorescence spectrometer. The peak absorption
wavelengths for PFO and SY were 384nm and
440nm respectively. Peak emission wavelengths
were 420nm for PFO and 590nm for SY.
Photoluminescent Quantum Yield (PLQY) was
measured with a Hamamatsu Photonics C9920-02
integrating sphere with excitation wavelengths of
384nm for PFO and 440nm for SY. The PLQYs of
the polymers were measured to be 46% for PFO, and
40% for SY.
2.2 Hardware
The excitation LED for each polymer was chosen to
match the absorption peak as closely as possible,
with a Royal Blue LUXEON LED from Philips
selected with a central wavelength of 447.5nm for
SY and a 370nm-centred LED from Thorlabs for the
PFO excitation. The appropriate LED was aligned
with the sample and detector as shown in Figure 2:
the excitation light was filtered using a high-pass
A Portable, Low-cost System for Optical Explosive Detection based on a CMOS Camera
133
filter, and a neutral density filter of optical density 1
or 2 was inserted to reduce any saturation effects on
the CMOS camera if necessary. The CMOS camera
module from Raspberry Pi has 5 megapixels and a
fixed focus.
Figure 2: Hardware set-up schematic. A Royal Blue
LUXEON LED excites the Super Yellow film, and the
High-Pass filter removes the source light. The emitted
light is collected by the camera and processed on-board
the Raspberry Pi platform.
2.3 Experimental Set-up
The polymer-coated slide was inserted into an in-
house designed airtight chamber with gas flow inlet
and outlet, and quartz windows on opposing sides
for light excitation and emission. Powdered
dinitrobenzene (Sigma Aldrich) was placed in a
glass tube with a nitrogen line flowing through the
powder, then via control valves leading to the
chamber inlet. To make a measurement, the nitrogen
flow was turned on and the luminescent emission
monitored in real time with the camera. The images
were taken at 20 second intervals. Measurements
were conducted at room temperature.
2.4 Image Processing
The image processing was performed in Python
using open source libraries from Raspberry Pi.
Briefly, the system works by taking the original
image, prior to binarization, i.e assigning “light” and
“dark” regions to the image. Subsequently automatic
circle detection defined the emission area, and
finally automatic Region-of-Interest selection is used
to define the monitored signal. The process is
illustrated in Figure 3, with the decision-making
flow-chart shown in Figure 4.
Figure 4: Flowchart for image processing.
Figure 3: Image Processing with an illuminated Super Yellow film.
PHOTOPTICS 2016 - 4th International Conference on Photonics, Optics and Laser Technology
134
3 RESULTS
Figure 5 shows the quenching profile of the Super
Yellow film to DNB vapour at a vapour pressure of
approximately 30 ppb(Ewing et al., 2013). A clear
and measurable decrease in luminescent intensity is
shown, with the quenching effect apparent after the
first 20 seconds. The intensity decreases by
approximately 6% in 200 seconds.
Figure 5: Quenching profile of Super Yellow exposed to
DNB vapour in nitrogen.
The response of PFO to DNB is shown in Figure
6. The quenching response is slightly lower than that
of SY at around 4.5%.
Figure 6: Quenching profile of PFO exposed to DNB
vapour in nitrogen.
These results indicate that the CMOS camera
system is a suitable method for optical detection of
ppb-level explosive vapours. The inherent difference
between the two polymers’ response to the DNB
vapour suggests that both could be monitored
simultaneously, leading to selective measurements
for identification of explosive vapours. The system
reported here is lightweight, inexpensive and low-
cost in addition to its sensitivity, so further work is
ongoing to ruggedize the system for field trials,
build a library of polymer-vapour response curves,
and investigate simultaneous multi-analyte sensing.
4 CONCLUSIONS
An optical sensing system was developed using an
off-the-shelf Raspberry Pi processing platform in
conjunction with a CMOS camera, using the
conjugated polymers Super Yellow and Polyfluorene
as the explosive vapour-sensitive sensor films. The
quenching effect of the vapour on the polymers’
luminescence was successfully monitored by the
camera and profiles generated over around 200
seconds of vapour adsorption to the polymer surface.
The system shows promise to be used as a portable,
user-friendly explosive vapour sensor in
humanitarian demining, and in this be useful
complementing other methods such as metal
detectors to more thoroughly detect and identify
buried landmines. Improvements and further work to
the system have been identified towards
development of a robust, highly-sensitive
instrument.
ACKNOWLEDGEMENTS
This project has received funding from the European
Union’s Seventh Framework Programme for
research, technological development and
demonstration under agreement no 284747.
REFERENCES
Bitella, G., Rossi, R., Bochicchio, R., Perniola, M. &
Amato, M. 2014. A Novel Low-Cost Open-Hardware
Platform for Monitoring Soil Water Content and
Multiple Soil-Air-Vegetation Parameters. Sensors, 14,
19639-19659.
Comina, G., Suska, A. & Filippini, D. 2015. Autonomous
Chemical Sensing Interface for Universal Cell Phone
Readout. Angewandte Chemie-International Edition,
54, 8708-8712.
Ewing, R. G., Waltman, M. J., Atkinson, D. A., Grate, J.
W. & Hotchkiss, P. J. 2013. The Vapor Pressures of
Explosives. Trac-Trends in Analytical Chemistry, 42,
35-48.
Leeuw, T., Boss, E. S. & Wright, D. L. 2013. In Situ
Measurements of Phytoplankton Fluorescence using
Low Cost Electronics. Sensors, 13, 7872-7883.
Narayanan, A., Varnavski, O. P., Swager, T. M. &
Goodson, T., Iii 2008. Multiphoton Fluorescence
0 50 100 150 200
92
94
96
98
100
92
94
96
98
100
Normalised Signal (a.u.)
t
(
s
)
0 50 100 150 200
95
96
97
98
99
100
95
96
97
98
99
100
Normalised Signal (a.u.)
t (s)
A Portable, Low-cost System for Optical Explosive Detection based on a CMOS Camera
135
Quenching of Conjugated Polymers for Tnt Detection.
Journal of Physical Chemistry C, 112, 881-884.
Newnham, P. & Daniels, D. 2001. The Market for
Advanced Humanitarian Mine Detectors. Conference
on Detection and Remediation Technologies for Mines
And Minelike Targets Vi, 2001.
Apr 16-20 2001 Orlando, Fl. 1213-1224.
Porritt, F., Shapiro, M., Waggoner, P., Mitchell, E.,
Thomson, T., Nicklin, S. & Kacelnik, A. 2015.
Performance Decline by Search Dogs in Repetitive
Tasks, and Mitigation Strategies. Applied Animal
Behaviour Science, 166, 112-122.
Thomas, S. W., Iii, Joly, G. D. & Swager, T. M. 2007.
Chemical Sensors based on Amplifying Fluorescent
Conjugated Polymers. Chemical Reviews, 107, 1339-
1386.
Tonacci, A., Corda, D., Tartarisco, G., Pioggia, G. &
Domenici, C. 2015. A Smart Sensor System for
Detecting Hydrocarbon Volatile Organic Compounds
in Sea Water. Clean-Soil Air Water, 43, 147-152.
Wang, Y., Mckeown, N. B., Msayib, K. J., Turnbull, G. A.
& Samuel, I. D. W. 2011a. Laser Chemosensor with
Rapid Responsivity and Inherent Memory based on a
Polymer of Intrinsic Microporosity. Sensors, 11, 2478-
2487.
Wang, Y., Turnbull, G. A. & Samuel, I. D. W. 2011b.
Conjugated Polymer Sensors for Explosive Vapor
Detection. Organic Semiconductors in Sensors and
Bioelectronics Iv, 8118.
Wang, Y., Yang, Y., Turnbull, G. A. & Samuel, I. D. W.
2012. Explosive Sensing Using Polymer Lasers.
Molecular Crystals and Liquid Crystals, 554, 103-110.
PHOTOPTICS 2016 - 4th International Conference on Photonics, Optics and Laser Technology
136