See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233864803

Electro-optical system for gunshot detection - Analysis, concept, and

performance

Article in Proceedings of SPIE - The International Society for Optical Engineering · June 2011
DOI: 10.1117/12.900968

CITATIONS READS
4 4,659

6 authors, including:

Mariusz Kastek Grzegorz Bieszczad

Military University of Technology Military University of Technology
123 PUBLICATIONS 1,000 CITATIONS 58 PUBLICATIONS 370 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mariusz Kastek on 30 May 2014.

The user has requested enhancement of the downloaded file.

 Electro-optical system for gunshot detection
– analysis, concept and performance
M. Kastek*, R. Dulski, H. Madura, P. Trzaskawka, G. Bieszczad, T. Sosnowski
Institute of Optoelectronics, Military University of Technology,
ul. Gen. Sylwestra Kaliskiego 2, 00-908 Warsaw, Poland

ABSTRACT

The paper discusses technical possibilities to build an effective electro-optical sensor unit for sniper detection using
infrared cameras. This unit, comprising of thermal and daylight cameras, can operate as a standalone device but its
primary application is a multi-sensor sniper and shot detection system. At first, the analysis was presented of three
distinguished phases of sniper activity: before, during and after the shot. On the basis of experimental data the
parameters defining the relevant sniper signatures were determined which are essential in assessing the capability of
infrared camera to detect sniper activity. A sniper body and muzzle flash were analyzed as targets and the descriptions of
phenomena which make it possible to detect sniper activities in infrared spectra as well as analysis of physical limitations
were performed. The analyzed infrared systems were simulated using NVTherm software. The calculations for several
cameras, equipped with different lenses and detector types were performed. The simulation of detection ranges was
performed for the selected scenarios of sniper detection tasks. After the analysis of simulation results, the technical
specifications of infrared sniper detection system were discussed, required to provide assumed detection range. Finally
the infrared camera setup was proposed which can detected sniper from 1000 meters range.
Keywords: sniper detection, muzzle flash detection, IR system for sniper detection, sniper signature

1. INTRODUCTION
The nature of current military conflicts differs significantly from the full-scale wars of the past. In most situations the
armies of developed countries are nowadays involved in asymmetric warfare. In such cases the sheer military power
superiority of a modern army cannot be fully utilized against guerilla forces, especially in the urban environment.
Countering this new threat will have to include the expansion of the technological advantage that modern armies already
have. One example where technology plays an important role is counter-sniper systems, which can be extremely useful
in urban guerilla warfare. In such military operations the importance of counter-sniper activity cannot be overestimated.
For example, data indicates that small arms fire accounted for 19.7% of American fatalities in April of 2006 while in
October that number had increased to 43%. This caused the urgent need do deploy effective anti-sniper systems and
eventually over 700 units have been sent to Iraq and Afghanistan.
Counter-sniper systems can use several different physical phenomena related to the shot or the weapon itself, such as
acoustic, visual, or electromagnetic signals. Initially acoustic signals, such as the muzzle blast and the ballistic
shockwave, provided the easiest and most accurate way to detect shots, and hence, the majority of existing counter-
sniper systems use them as the primary information source.
Modern, multi-sensor systems utilize also infrared and visual signatures of the shot (muzzle flash), infrared, visual or
radar signatures of the flying bullet and laser radiation reflected from the sniper’s scopes. In the following paper the
possibility of sniper detection using signals occurring in the infrared range will be discussed.
The phenomena detected in infrared spectra are muzzle flash and thermal signatures of the bullet in flight and of sniper
body. Muzzle flash is an infrared signature associated with the ejection of the bullet from the sniper’s rifle. The muzzle
flash can be detected with IR sensors out to a kilometer or more, but the sensors must have clear line of sight to the
weapon, and the flash can be suppressed. The thermal signature of the bullet in flight can be detected with infrared
sensors out to several kilometers in range. Since the bullet is much hotter than “room temperature,” it is detected most
effectively in the medium-wave infrared (MWIR) band, with wavelength between 3 and 5 μm.
*[email protected]; phone +4822 6839383; fax +4822 6668950; www.wat.edu.pl
The object of detecting signatures of the bullet in flight is to estimate the bullet’s trajectory and backtrack it to find the
location of the sniper [1, 2, 3].
100
90

80

70

Intensity normalize
60

50

40

30

20

10

0
1 3 5 7 9
λ [μm]

Figure 1. The spectral distribution of infrared radiation from secondary flash (caliber 0.5 gun, barrel length 36 in) [4].
The phases of a shot event with corresponding temperature and pressure changes are shown above in Fig.1. The analysis
of the shot detection in different spectral bands revealed, that very little information is available on the ultraviolet
radiation of gun flash. It was found that possibly only in 404.4 - 404.7 nm range the strong line of atomic potassium can
be detected by UV sensor. However, during both computer simulations and real measurements no UV radiation was
detected during shot recordings from various types of weapons (the experiment details are presented below). The spectral
shot characteristics presented in Fig. 4. clearly indicate, that most of the muzzle flash energy falls into near infrared
range, which is decisive for shot detection. Muzzle flashes generally peak in the 2.5 - 3.5 µm region (H2O) following
800 - 1400 K blackbody curves corresponding to water vapor lines and have other peaks at 2 microns and 4.7 microns.
As the target range increases, the radiances in the 1.7 - 2 µm, the 2.4 - 3.4 µm, and 4.0 - 4.5 µm range decrease rapidly
[3, 4, 5]. Three stages can be distinguished by analyzing sniper activity on a battlefield:
• stage I – preparations for the shot,
• stage II – the shot,
• stage III – withdrawal after the shot.
During each of the above stages different, specific phenomena can be observed (Fig. 2). In Stage I and III a skilled sniper
uses his abilities to camouflage and takes advantage of terrain properties to remain undetected in both visible and
infrared spectral ranges. Analysis of the possibilities of sniper detection in a forested, mountain or urban area (most
common in present asymmetric conflicts) shows, that camouflage sometimes is quite effective. Thermal cameras have to
detect small temperature differences between sniper and background by analyzing the temperature distribution over
observed scenery. As a result very sensitive cameras hale to be used and, additionally, they should be equipped with
automatic search and track modules. Searching for and tracking sniper-like targets is a complicated task to perform.

Figure 2. The sniper in firing position (visible and infrared images).

On the basis of the measurement data obtained during laboratory and field tests at Military University of Technology and
using data presented in [1 - 6] it was possible to determine the size and temperature of the gas cloud emerging from the
barrel during the shot. Additionally, the energy levels emitted in two analyzed spectral bands (MWIR and LWIR) can be
compared. The analysis of the recorded thermal images leads to the determination of thermal properties of the shot event
and the sniper body. Sniper body thermal signature results from the difference, or contrast, between the body and its
immediate background. Imaging sensors see internal target detail and external shape detail. Therefore, target signatures
are defined by their pattern features. Those features are unique only to the extent that their proprieties differ from those
in the background. Thus, resolved sniper body signatures depend on background intensity mean values as well as on
clutter intensity variations on a size scale comparable to internal target detail [8]. The results of example shot recording
made by measurement thermal camera during field tests are presented in Fig. 3. Raw measurement data were analyzed
using AltaIR and ThermaCAM Researcher software.

Figure 3. The sniper body thermal signature registered during field tests (IR image on the left and VIS image on the right).

2. ANALYSIS OF LIMITATIONS OF INFRARED CAMERA SYSTEMS

The essential properties of a thermal camera that influence its sniper detection capabilities are the parameters of optics
and focal plane array detector.
2.1 Optics limitations
Spatial resolution of the lens is limited by diffraction. This restriction is particularly important in the infrared spectral
range, in both first (3-5 μm) and second (8-12 μm) atmospheric window. The values defining the threshold spatial
resolution of the lens-array combo are: the size (diameter) of Airy disc D (which describes the best focused spot of light
that a perfect lens with a circular aperture can make, limited by the diffraction of light) and pixel size of a single detector
in an array (pixel pitch). In practical applications it is assumed, that using „reasonable” number of optical elements
yielding average transmission not less than 70% it is possible to construct the lens with F#=1 and aberrations corrected
up to the diffraction limit. It is further assumed, that aberrations are fully corrected for lenses with small (up to 100) field
of view. For wider angle lenses this requirement is not applied at the extreme borders of the field of view. It is required
that the lens does not limit the spatial resolution of the lens-array combo. It should be mentioned, however, that too small
pixel pitch may lead to such situation. In practice two conditions are used to assure proper parameters of the lens-array
combo:
• Airy disc diameter equals to the pixel size, with aberrations corrected at diffraction level. It is often used safe
condition,
• two diameters of Airy disc equal three pixel sizes.
Taking into account the second condition of the above ones, the minimal pixel sizes can be calculated for MWIR and
LWIR spectral ranges. The calculated values, for F# = 1 are:
wavelength λ pixel size a x a
4μm 6.51 x 6.51μm
10μm 16.27 x 16.27μm
Assuming F# = 1, it is then questionable to use smaller pixel sizes, as they don’t improve overall resolution. Bearing in
mind, that practical diameter of infrared lens is limited to about 150 mm, the instantaneous field of view (IFOV)
corresponding to threshold pixel pitch (for F#=1) can be calculated:
Δλ IFOV
8-12μm 0.11mrad
3-5μm 0.04mrad

The construction of the lens with long focal length (in order to increase spatial resolution) may be difficult and
expensive, (especially if F#=1 is to be retained), because huge optical elements made from IR-transmissive materials
have to be used. In an initial attempt to estimate the required parameters of a system, capable to achieve desired ranges,
the pure geometrical approach was used and the required values of IFOV were calculated. Those values depend solely on
geometrical relations between target size and distance. The total FOV depends on array size (in pixels). The required
spatial resolution for detection, recognition and identification (DRI) was calculated for common array sizes of 320x240,
640x480 and 1024x768 pixels.
Each target was represented by equivalent rectangle and thermal contrast. The critical dimension was calculated for all
targets, as square root of target area. This critical dimension was used to calculate required pixel size in target plane,
because, according to Johnson criteria, critical target dimension must be subtended by certain amount of pixels to assure
50% probability of detection (1,5 pixel), recognition (6 pixels) and identification (12 pixels) [9]. Knowing the expected
pixel size in the target plane it was then possible to calculate the parameters of the lens (FOV and focal length), for
certain array size and pixel pitch. For example, for detection case, the following lens parameters were calculated
concerning sniper body detection case up to 500m distance.
Table 1. Sniper body detection – lens parameters for 500m distance.
Array size FOV [0]
320 12,1
640 23,9
1024 37,3
Pixel pitch [μm] Focal length [mm]
35 53
25 38
17 26
7 11

The results indicate that it is possible to detect sniper presence at the distances more than 1000 meters. Unfortunately the
detection is not sufficient, because at least recognition level should be achieved, if not identification one. It is still
possible for pixel pitch of 17 μm or smaller. However for 1024x768 array operating in LWIR spectral range the focal
length of 354 mm is required. In MWIR range it is theoretically possible to reduce pixel pitch to 7 μm (diffraction limit),
resulting in required focal length of 145 mm. These geometrical limits may not be sufficient, as this analysis does not
include thermal contrast values and detector parameters. For muzzle flash detection case the following lens parameters
were obtained.
The results indicate that it is also possible to detect muzzle flash at the distances longer than 1000 meters. However, the
detection level may not be sufficient and the performance between detection and recognition is expected. Such ranges
can be achieved for pixel pitch of 25 μm or smaller. For example for 1024x768 array with 25 μm pixel pitch, operating
in LWIR spectral range the focal length of 318 mm is required. The application of 17 μm pixel pitch array reduces the
focal length to 216 mm, and in MWIR (theoretical 7 μm pixel pitch limit) the focal length of 90 mm is sufficient.
 Table 2. Muzzle flash detection – lens parameters for 500m distance.

Array size FOV [0]

320 21,6
640 41,8
1024 62,8
Pixel pitch [μm] Focal length [mm]
35 29
25 21
17 14
7 6

2.2 Detector limitations

Microbolometers are the fastest growing type of thermal detectors. They don’t require cooling, operate at room
temperature and they are also affordable. The main materials for microbolometer detectors are amorphous silicon (a-Si),
vanadium oxide (VOx), yttrium barium copper oxide (YBaCuO) and barium strontium titanate (BST). Two types of
microbolometer array are commercially available: a-Si and VOx-based devices. The advantage of amorphous silicon is
its direct ability to be monolithically integrated onto silicon CMOS readout integrated circuits using well mastered
deposition and etching technologies. On the other hand, VOx devices offer higher TCR (thermal coefficient of
resistance) value but also higher noise level than a-Si parts. Additionally it is more difficult to create VOx detectors
which raise the costs of VOx arrays [10]. Resolution of microbolometer arrays has also been increased over the years and
1024x1024 arrays (pixel pitch 17 μm) were presented in 2008 r. There are several types of cooled infrared detectors.
Among them, MCT, InSb and QWIP detectors are most common and widely used.
Mercury Cadmium Telluride (MCT) is an alloy of two semiconductors, mixed together in non-stoichiometric ratio x,
symbolically denoted as Hg1-xCdxTe.The band gap depends on the aforementioned x ratio, so the spectral response of a
MCT detector can be shaped up across the whole infrared range. Tailorable energy band gap over the 1–30µm range,
large optical coefficients that enable high quantum efficiency and favorable inherent recombination mechanisms that
lead to high operating temperature – all those features make HgCdTe alloy nearly ideal IR detector material.
Indium antimonide (InSb) is a chemical compound, which has a direct band gap (at 77K) of Eg = 0.23eV, which
corresponds to cut-off wavelength of λc ≈ 5.5mm. Resulting detectivity reaches the level of D* ≈ 1011 cmHz1/2W−1,
which makes them one of the most sensitive IR detectors for MWIR range (0.9 ¾ 5.0 µm). The detectors based on the
InSb binary direct band gap semiconductor exhibit a high quantum efficiency equivalent to MCT. The disadvantage of
InSb is that band gap is reduced as temperature increases, which increases strongly the dark current and limits the
operating temperature to around 80K.
QWIP detectors are made mainly of the lattice matched GaAs/AlGaAs material system. Single structure consists of
GaAs, InGaAs and AlGaAs layers. QWIP detectors are, by nature, narrow-band devices, however by special design it is
possible to achieve wide-band properties. Technological capabilities allow making the arrays with more than 1M pixels.
Constant technology improvement results in bigger arrays and lower NETD values. Furthermore the biggest
disadvantage – very low operating temperature (initially below 50K) was significantly improved in the newest
generation of devices and current QWIP arrays are now on par with MCT counterparts.
2.3 Comparison between 3-5 μm and 8-12 μm spectral ranges
The sizes of recorded muzzle flashes are similar in both bands, but thermal contrast is much higher in MWIR (as
expected) (results of analysis are shown in Fig. 4). MWIR spectral range is generally better suited for the detection of
high temperature objects, but in case of muzzle flash the recorded apparent temperatures are not extremely high. As a
result, both MWIR and LWIR thermal cameras can detect muzzle flash from all considered angles of observation [2, 7].
2.4 Camera frame rate for muzzle flash detection
The duration of the muzzle blast varies with weapon type from single milliseconds to tens of milliseconds. Such short
duration implies the need for a fast camera to register the phenomenon and as a result frame rates up to 500 fps are
required. During experiments concerning the registration of thermal signatures the muzzle flash recordings were
performed by measurement cameras with frame rates ranging from 200 to 350 Hz. In every registration at least several
frames of muzzle flash were recorded. Even relatively slow uncooled microbolometer camera with frame rate 30 fps was
sometimes able to record single frame with muzzle flash image. For reliable detection of muzzle flash, however, frame
rates of hundreds of fps are required.

Figure 4. Muzzle flash images in two spectral bands:

MWIR 3-5μm (on the left) and LWIR 8-12μm (on the right).

3. RESULTS OF CALCULATIONS FOR THE DETECTION, RECOGNITION AND

IDENTIFICATION OF SNIPER
The thermal camera chosen for sniper detection must meet numerous requirements. The thermal camera should work in
the LWIR and/or MWIR spectral ranges and shall be equipped with anti-reflection coated lenses wide/narrow switchable
optics for target detection/identification. Due to environmental requirements the IR camera shall be mounted on a heavy-
duty pan/tilt drive and rated for outdoor installations. The pan drive shall be capable of +/-200° panning and the tilt
drive shall be capable of +/- 30° tilt. The thermal camera should also be stabilized due to narrow optics used for
recognition/identification task. The thermal camera shall include a Dynamic Digital Detail Enhancement (DDE) filter
which is a real-time DSP-based image enhancement function that optimizes the thermal image across all scene
conditions to provide video with the optimal contrast and low noise. It also shall utilize Non-Uniformity Correction
(NUC) which is a set of compensation factors for each pixel. The camera shall feature both White-Hot and Black-Hot
operating modes, where White-Hot mode (default) means warmer objects will be displayed in white or lighter shades
than cooler background areas while in the Black-Hot mode warmer images will be displayed as black or darker than the
background. Additional pseudo-color palettes shall be available using the application software provided with the camera.
An image recognition algorithm shall be implemented in the camera for automatic target recognition as well as remote
control and possibility to connect to other sensor for data fusion.
Table 3. Specifications of analyzed thermal cameras.
Sensor Type FPA, uncooled VOx microbolometer 320 x 240 FPA, cooled InSb 1024x768
Pixel size 38μm 17 μm
Spectral Range 7.5 μm to 13.5μm 3-5μm
Lens Focal Length 50mm 354mm
F# 1.7 2.24
Field-of-View 14° (H) x 10° (V) 2.8°x2.1°
The initial assumption was to obtain target discrimination range of 1500 meters (for both sniper body and muzzle flash),
just beyond the effective range of most small arms weapons. Two kinds of IR cameras were chosen for simulation and
analysis. First one with uncooled microbolometer FPA works in LWIR range and the second one with cooled detector
working in SWIR range. One of the analyzed cameras was PTZ-50 MS, commercially available unit by Flir Systems.
The second camera was high-resolution cooled device with 1024x768 FPA detector. Technical specifications of both
cameras are presented in Table 3. Range calculations were made using NVTherm software [11, 12], and the appropriate
results for the devices with selected technical specifications were compared. An example NVTherm calculations for
cooled InSb 1024x768 pixels thermal camera detecting sniper body and muzzle flash are presented in Fig. 5. Thermal
properties of a sniper body and muzzle flash were assumed on the basis of STANAG 4347 norm and experimental
examination performed at a test field.
1.0 1.0

0.8 0.8
Probability

Probability
0.6 Detection 0.6 Detection

0.4 0.4

0.2 0.2

0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Range [ km ] Range [ km ]
Figure 5. Probability of detection as a function of distance in km for sniper body:
(a) and muzzle flash (b). Results for cooled InSb 1024x768 pixels thermal camera.
The calculations for other cameras, equipped with different lenses and detector types were conducted in a similar way.
Several different types of thermal cameras were analyzed and the example results of range calculations for four
representative camera types are presented in Tab. 4.
Table 4. Range calculations for analyzed thermal cameras.
Type
Parameter Value Target Det. Recogn. Ident. Remarks
of camera
LWIR pixel pitch 17μm sniper
1024x768 5210m 1640m 870m input aperture
lens diameter 160mm body diameter
NEDT=85mK limited to 160
F# 2.2 muzzle
(μbolometer 6980m 2090m 1170m mm
cooled) FOV 2.8°x2.1° flash
pixel pitch 17μm sniper increased
LWIR 8750m 2640m 1380m
lens diameter 160mm body detector
NEDT=10mK sensitivity
1024x768 F# 2.2 muzzle (theoretical
(uncooled) 8610m 2480m 1310m case)
FOV 2.8°x2.1° flash
pixel pitch 17μm sniper
MWIR 7230m 7340m 1250m standard
1024x768 lens diameter 160mm body
cooled
D*=1x1011 F# 2.2 muzzle
(Insb cooled) 8690m 2680m 1410m technology
FOV 2.8°x2.1° flash
By analyzing the above tables one can reveal that standard high quality security camera cannot meet the assumed range
requirements for sniper body detection and muzzle flash detection.
Uncooled thermal camera with the following parameters: array size 1024x768 pixels, FOV 2.8°x2.1° theoretically meets
such requirements. However, the assumed F-number =1 is well beyond standard technology capabilities because of huge
lens diameter. Otherwise aperture limit significantly reduces the DRI ranges. Even detector NETD value of 10 mK
(currently unavailable) is not sufficient to achieve the desired ranges of sniper detection. It seems that thermal camera
with uncooled detector working in LWIR spectral range can be used to detect muzzle flash only. The thermal camera
with cooled detector with typical peak detectivity D* has better DRI ranges than its uncooled counterpart, but meets the
assumed range requirements only in case of muzzle flash detection. Three-times increase in detectivity value (Peak D*)
allows to achieve the desired ranges of sniper detection. The availability of such detector is to be determined, but it
should be mentioned that the lens parameters (focal length 354 mm, lens diameter 160 mm) make such optics difficult to
manufacture and rather expensive. The application of cooled detector with theoretically possible pixel pitch of 7μm
meets sniper detection requirements both sniper body and muzzle flash from a distance up to 1500 m (identification
case). The reduction of pixel pitch influences also lens parameters and the focal length, but it is purely theoretical case as
no such detectors are currently available.
The calculations for other cameras, equipped with different lenses and detector types were conducted in a similar way.
Several different types of thermal cameras were analyzed. On the basis of the collected data the optimal parameters of a
thermal camera for shot detection purposes were determined, which allow for the desired target discrimination ranges.
In order to cover the threat localization up to 360° in azimuth it is necessary to apply the area scanning (for single IR
camera system). It can be seen that different FOV values are required for muzzle flash and sniper detection. In a single
device it can be achieved by using zoom lens, which are currently available (and affordable) for infrared spectral range.
Sample configuration of such system is presented in Fig. 6.

360o

ZOOM
± 30o

FOV NARROW
FOV WIDE

Figure 6. Camera with selectable field of view.

The camera, with two selectable FOV (22°x16.7° & 3.6°x2.7) is used to detect sniper presence and muzzle flash. The
analysis results show that such task can be accomplished by quite standard lens and array camera e.g. cooled LWIR
device with array size of 384x288 pixels. However, MWIR range is generally more effective for the detection of muzzle
flash. Additionally fast frame rates (hundreds of fps) are required which in this case will be the limiting factor. Upon
detection, the second camera will be automatically pointed in the direction of the potential threat and final decision will
be made. Considering present technology level it can be stated that detection ranges more than 3000 meters are quite real
to achieve, but the recognition to is about 600 meters.

4. LABORATORY MODEL OF AN INFRARED CAMERA

The model was created using LWIR HgCdTe detector manufactured by Sofradir, France. It is offered under Venus brand
and it is 384x288 pixel cooled FPA device witch 25 μm pixel pitch. Cooling is provided by integrated K508 Sterling-
type cooler (manufactured by Rico company, Israel). The applied Venus LW detector is a LWIR device, with peak
sensitivity at 8.2 μm. The spectral range is defined by germanium input window with AR layer and cut-off filter at 9.5
μm and by cooled diaphragm with optical cut-on filter at 7.5 μm. Signal conversion module is based on EASYLINK
board, which main task is to convert analogue detector signals into 14-bit digital data. Block diagram of the constructed
camera, including its basic functional modules is presented in Fig. 7.
 LENS
CONTROL

VIDEO DATA IMAGE DIGITAL IMAGE READ-OUT COOLED MOTORIZED

TRANSMISSION ANALYSIS PROCESSING CIRCUIT FPA ZOOM LENS

Figure 7. Simplified block diagram of IR camera.

Read-out circuit is based on commercial-off-the-shelf EASYLINK board, which has internal clock generator allowing
for broad selection of single detector read-out frequency: 10 MHz, 5 MHz, 2.5 MHz or 1.25 MHz [16]. The application
of cooled FPA implies the use of precise temperature control circuits. The FPA temperature greatly influences its
sensitivity so it is very important to keep the temperature at optimal level. Another key factor to stable camera
performance is the quality of supply voltages.

Figure. 8. Laboratory model of the IR camera.

Output video data are calculated from FPA signals (converted into digital data) and correction coefficients stored in
internal memory. Digital image data stream is converted into output video signal according to CCIR/PAL standard by a
special D/A converter module [16, 17]. The same electronic module generates also horizontal and vertical
synchronization signals. The output video can be displayed on various types of displays, such as CRT, LCD and OLED.
The laboratory model of IR camera was fitted with RONIT zoom lens by Temmek Optics. Its 25-150 mm zoom range
corresponds to FOV range from 22°x16.7° (wide) to 3.6°x2.7° (narrow). The model was tested on a laboratory stand
with optical collimator, reference IR source and bar tests. During the tests the basic parameters of the camera were
determined, such as NETD and MRT and SRF characteristics. Further tests are planned, including field tests for muzzle
flash and sniper body detection in various environmental conditions to verify the design assumptions.

5. CONCLUSIONS
Sniper detection suddenly become one of the most challenging tasks that opto-electronic systems for battle
reconnaissance have to cope with. The sniper detection before he takes the shot is of primary importance. This can be
achieved by the detection of thermal contrast between (usually camouflaged) sniper body and background. It is not an
easy task and the technical requirements for the effective systems with suitable thermal and spatial resolutions are quite
critical. The presented analysis revealed that microbolometer-based thermal cameras does not have the capability to
achieve long detection ranges, often needed in counter-sniper operations. The sniper detection at the ranges of a
kilometer and more require the best of currently available cooled focal plane array detectors and top quality optics. The
initial parameters of the camera for sniper and shot detection were determined on the basis of laboratory and field
measurements of thermal properties of both muzzle flash and sniper body. Further calculations using NVtherm software
revealed the necessary camera parameters to obtain sniper detection range of 1500 meters. On the basis of measurements
and numerical simulations the camera model was constructed with selectable field of view, optimized for sniper body
and muzzle flash detection at the desired range. This model was tested under laboratory conditions and, in the next step,
the field tests will be performed, as final verification of expected ranges of sniper and muzzle flash detection.
 The result presented in paper are supported by realization of the Project co-financed by the European Regional
Development Fund within the framework of the 2. priority axis of the Innovative Economy Operational Programme,
2007-2013, submeasure 2.1. "The development of centres with high research potential". Contract no. POIG.02.01.00-14-
095/09.

REFERENCES

[1] Spera T. J. and Figler B. D., “Uncooled infrared sensors for an integrated sniper location system”, Proc. SPIE
Vol. 2938, pp. 326-339, (1997).
[2] Csanadi C. J., Edwards G. D., Hintz T. M. and Tong R. M., “Multispectral signature analysis measurements of
selected sniper rifles and small arms”, Proc. SPIE, Vol. 2938, 288, (1997).
[3] Richards A. A. and Risdall D. M., “Passive Thermal Imaging of Bullets in Flight”, Proc. SPIE Vol. 5405,
pp.258-263, (2004).
[4] Klingenberg G. and Heimerl J. M., “Gun Muzzle Blast and Flash”, Progress in Aeronautics and Astronautics,
vol. 139, (1992).
[5] Law D., “Multi-spectral Signature Analysis Measurements of Selected Sniper Rifles and Small Arms”, SPIE,
Vol. 2938, 8 (1997).
[6] Grant B. G. and Hardy D. T., “Muzzle Flash Issues Related to the Waco FLIR Analysis”, in Proc. SPIE, vol.
4370, pp. 314-324, (2001).
[7] Kastek M., Dulski R., Trzaskawka P., Bieszczad G., “Measurement of sniper infrared signatures”, Proc. SPIE
Vol. 7481, 748113, (2009).
[8] Dulski R., Madura H., Piątkowski T., Sosnowski T., “Analysis of a thermal scene using computer simulations”,
Infrared Physics & Technology, No 49, p. 257-260, (2007).
[9] Holst G. C., “Electro-optical Imaging System Performance”, JCD Publishing, Winter Park, FL, (l995).
[10] Callen C., Goodman J., “Sensors to Support the Soldier”, The Jason Study, The MITRE Corporation, McLean,
VA, (2006).
[11] FLlR92 Thermal Imaging Systems Performance Model, User's Guide, NVESD document UG5008993, Fort
Belvoir, VA (1993).
[12] Ratches J. A., Lawson W. R., Obert L. P., Bergemann R. J., Cassidy T. W., and Swenson J. M., “Night Vision
laboratory Static Performance Model for Thermal Viewing Systems”, ECOM Report ECOM-7043, pg. 2, Fort
Monmouth, NJ (1975).
[13] Kastek, M., Dulski, R., Trzaskawka, P., “Bieszczad, G., Sniper detection using infrared camera: technical
possibilities and limitations”, Proc. SPIE 7666, 76662E (2010).
[14] Dulski, R., Kastek, M., Bieszczad, G., Trzaskawka, P. and Ciurapiński, W., “Data fusion used in multispectral
system for critical protection”, WIT Transactions on The Built Enviroment, 108, pp.165-173 (2009).
[15] Włodarski, M., Kopczyński, K., Kaliszewski, M., Kwaśny, M., Mularczyk-Oliwa, M. and Kastek, M.,
A”pplication of advanced optical methods for classification of air contaminants”, WIT Transactions on Ecology
and the Enviroment, 123, pp.237-247 (2009).
[16] Bieszczad, G., Sosnowski, T., Madura, H., Kastek, M., Barela, J., “Adaptable infrared image processing module
implemented in FPGA,” Proceedings of SPIE Vol. 7660, 76603Z (2010).
[17] Sosnowski, T., Bieszczad, G., Madura, H., Kastek, M., Firmanty, K., “The calibration stand for thermal camera
module with infrared focal plane array”, Proc. SPIE Vol. 7660, 76603Y (2010).
[18] Bieszczad G., Sosnowski T., Madura H., Kastek M., Barela J., “Adaptable infrared image processing module
implemented in FPGA”, Proc. SPIE Vol. 7660, 76603Z (2010).
[19] Trzaskawka P., Dulski R., Kastek M., “Concept of electro-optical sensor module for sniper detection system”,
Proc. SPIE Vol. 7834, 783414 (2010).
[20] Kastek M., Dulski R., Trzaskawka P., Sosnowski T, and Madura H., “Concept of infrared sensor module for
sniper detection system”, IRMMW-THz 2010 - 35th International Conference on Infrared, Millimeter and THz
Waves, Conference Guide Article number 5612447 (2010).

View publication stats