US8809787

US008809.
787B2
(12) United States Patent (10) Patent No.: US 8,809,787 B2

Tidhar (45) Date of Patent: Aug. 19, 2014
(54) GUNSHOT DETECTION SYSTEMAND 3,936,822 A 2/1976 Hirschberg
METHOD 5,001,348 A 3, 1991 Dirscherlet al.
5,455,868 A 10/1995 Sergent et al.
5,686,889 A 11, 1997 Hillis
(71) Applicant: Elta Systems Ltd., Ashdod (IL) 5,703,321. A 12/1997 Feierlein et al.
5,751,215 A 5/1998 Hall, Jr.
(72) Inventor: Gil Tidhar, Modiin (IL) 5,912,862 A 6/1999 Gustavsen et al.
5,973,998 A 10, 1999 Showen et al.
(73) Assignee: Elta Systems Ltd., Ashdod (IL) 6,178,141 B1 1/2001 Duckworth et al.
6,496,593 B1 12/2002 Krone, Jr. et al.
-r - 6,781,127 B1 8/2004 Wolff et al.
(*) Notice: Subject to any disclaimer, the term of this 6,969,856 B1 1 1/2005 Hillenbrand et al.
patent is extended or adjusted under 35
U.S.C. 154(b) by 0 days. (Continued)
(21) Appl. No.: 14/027,844 FOREIGN PATENT DOCUMENTS
1-1. EP 1112459 T 2001
(22) Filed: Sep. 16, 2013 EP 1811315 A2 7, 2007
GB 2280S63 A 2, 1995
(65) Prior Publication Data WO O223227 A2 3, 2002
US 2014/O184806A1 Jul. 3, 2014 OTHER PUBLICATIONS
Related U.S. Application Data Israel Patent Application No. 118619 filed Jun. 11, 1996 entitled
(63) Continuation of application No. 12/841.289, filed on “Dual Sensor Surveillance System”.
Jul. 22, 2010, now abandoned, which is a continuation (Continued)
of application No. PCT/IL2008/000105, filed on Jan.
23, 2008.
Primaryy Examiner — Marcus Taningco
9.
(51) Int. Cl. (74) Attorney, Agent, or Firm — Browdy and Neimark,
HOIL 27/4 (2006.01) PLLC
(52) U.S. Cl.
USPC ..................................................... 250/339.02
(58) Field of Classification Search (57) ABSTRACT
CPC HO4N 5/332 s GO6K9/OO771 s GO6K9/3233 s A device and a method for use in detection of a muzzle flash
USPFoll 27/14; H01L 27/14643; Holy55: event is described. The device can include a Photo Detector
S licati - - - - - file? - - - - - - - - - - - - -let - - - - - - - - - - -hhi t Array (PDA), sensitive in at least a portion of the NIR and
ee appl1cauon Ille Ior complete searcn n1Story. SWIR spectrum, and a filter of electromagnetic radiation
(56) References Cited selectively passing in this portion a spectral range of low
atmospheric transmission, the PDA has an integration time
U.S. PATENT DOCUMENTS shorter than a duration of the muzzle flash event.
3,699,341 A 10/1972 Quillinan et al.

3,931,521 A 1/1976 Cinzori 19 Claims, 9 Drawing Sheets
it is
2.
L.. 8: 5,As 20Aassis
&
ax- s
- Processor'
x.Y sParaitei
S-H pdrp Processing
- - . .a Set s 46
w N irit a s
'- 8 w w w w x - ww.
Niuzzie
US 8,809,787 B2
Page 2
(56) References Cited Ando et al., “Correlation Image Sensor: Two-dimensional Matched
Detection of Amplitude-modulated Light', IEEE Transactions on
U.S. PATENT DOCUMENTS Electron Devices, vol. 50, No. 10. Oct. 2003, p. 2059.
V. Douence, et al., "Hybrid Image Sensor with Multiple On-chip
7,202,794 B2 4/2007 Huseynov et al. Frame Storage for Ultra High Speed Imaging”. Proc. of SPIE, vol.
2006/0021498 A1 2/2006 Moroz et al. ................. 89.41.06 5580, 2005.
2007/O125951 A1* 6, 2007 Snider et al. ............. 250,363.03 “Laser-based Sniper detection system unveiled for army robot'. Laser
2007. O158638 A1 7/2007 Perera et al. Focus World, Oct. 3, 2005, Boston, MA. Burlington MA.
OTHER PUBLICATIONS A. Richards et al., “Passive Thermal Imaging of Bullets in Flight'.
Proc. of SPIE, vol. 5405 (SPIE, Bellingham WA, 2004), p. 258-263.
A. Vick et al., “Aerospace Operations in Urban Environments: M. Kastek et al., “Sniper detection using infrared camera technical
possibilities and limitations”. Proc. of SPIE, vol. 7666.2010.
Exploring New Concepts”. RAND Corporation monograph/report Settles, “The Penn state full-scale sch I ieren system' 11th Interna
MR-1187-AF, pp. 213-215, 2002. tional Symposium on Flow Visualization, Aug. 9-12, 2004 University
A. Richards, "Applications for high-speed infrared imaging'. of North Dame. Notre Dame, Indiana, USA, p. 1-12.
Abstract. Proc. SPIE vol. 5580, p. 137-145, 2005. Kagan, “Lessons from the battle of Grozny, 1994-1995”, “Conven
A. Richards, “Infrared Camera Measures Bullet Heating”. Advanced tional and unconventional warfare since 1945”,US Military Acad
Imaging Magazine, Mar. 2004. emy, Cadet Sean Mccafferty OO.co E4 West Point, NY May 1, 2000.
G. Settles et al., “Full-scale high-speed Schlieren Imaging of Explo R. Dlugosz et al..."High-precision analogue peak detector for X-ray
sions and Gunshots.” SPIE paper 5580-174. Proc. 26th Inti. Conf. on imaging applications'. Electronics Letters, vol. 43, No. 8, p. 440,
High-Speed Photography and Photonics, Alexandria, USA, 2004. Apr. 12, 2007.
G. Settles et ai..."Fuii-Scale High-Speed “Edgerton' Retroreflective Burton et al. “Improved Integrated Sniper Location System' Part of
Shadowgraphy of Explosions and Gunshots”, 5th Pacific Symposium SPIE Conference on Sensors, c31, Information, and Training Tech
on Flow Visualization and Image Processing, 2005, Australia. nologies for Law Enforcement, Boston Massachusetts, vol. 3577
Pauliet al., “Infrared Detection and Geolocation of Gunfire and Ord Nov. 1998, vol. 3577, p. D 243-252.
nance Events From Ground and Air Platforms”, 6 NATO Symposium A. White, “Incoming fire: Sniper-detection systems Support counter
on "Systems, Concepts and Integration (SCI) Methods and Technolo insurgency operations', I DR, Oct. 24, 2006 p. 1-10.
gies for Defence Against D Terrorism” London, UK, 2004. G.L Duckworth et al., “Fixed and wearable acoustic counter-Sniper
M. Ertem, "An Acoustic Sensor for the VIPERInfrared Sniper Detec systems for law enforcement Sensors, c31, Information, and Train
tion System'. Military Sensing Symposia, IRIS Specialty Group on ing Technologies for law Enforcement SPIE Proceedings vol. 3577.
Battlefield Acoustics and Seismics, 1999. Nov. 3-5, 1998, Boston, MA.
T. Spera et al., “Uncooled infrared sensors for an integrated Sniper M. Zahler et al. “SWAD-Small arms fire Warning and Direction
location system'. SPIE Proceedings, vol. 2938, pp. 326-339. finding System—a passive IP concept” Proc. of SPIE vol. 6542,
A. Fish et al., “An APSWith 2-D Winner-Take-All Selection Employ 654237, (2007).
ing Adaptive Spatial Filtering and False Alarm Reduction', IEEE Hamamatsu, “InGaAs Linear Image Sensor. Image sensors, May
Transactions on Electron Devices, vol. 50(1). Jan. 2003. 2005, pp. 1-4, XP0025 10267.
J.Ohta et al., “An Image Sensor with an In-Pixel Demodulation International Search Report, mailed Jan. 27, 2009 from PCTIIL2008/
Function for Detecting the Intensity of a Modulated Light Signal”, 000105, filed Jan. 23, 2008.
IEEE Transactions on Electron Devices, vol. 50(1), Jan. 2003. Nelson, R.J. “Integrated Hostile Fire Indication Sensor'. Solid State
M.C. Pitter et al., “Dual-phase synchronous light detection with Scientific Corporation, NAVAIRPublic release 09-134, Jun. 2, 2008,
64'64 CMOS modulated light camera'. Abstract. Electronics Letters, available at: http://www.virtuallacquisitionshowcase.com/document/
vol. 40(22), Oct. 2004. 1040/briefing.
M.C. Pitter et al., “Phase-Sensitive CMOS photo-circuit array for Moroz Saetal "Airborne Deployment of and Recent Improvements
modulated thermoreflectance measurements'. Electronics Letters, to the Viper Counter Sniper System.” Naval Research Laboratory,
Sep. 2003, vol. 39(18) Abstract. Washington, DC 1999, available at: www.dtic.mil/dticitrfulltext/u2/
S. Bourquin et al., “Two-dimensional Smart detector array for a390309.pdf.
interferometric applications'. Abstract. Electronics Letters, Jul.
2001, vol. 37(15). * cited by examiner
U.S. Patent Aug. 19, 2014 Sheet 1 of 9 US 8,809,787 B2
Reta\; i, AYAN )
us - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
{{ISSNS - &£S. sGNALs *RS). -ie A {
rt re
SSECTING XE (S), widoid KEL, iO,
r-------
R
{r \ {}^ visis - AS S
3.
is as a SSS
Yr r or a r Yr as r rar Yr rar at a n rar as r a war vs. r. Ir at r rar an a rar in a as ran r is as ran r is a room a s r.
3.
R
x
i
:
^{& -i-. SEE CES XSS
- it is -- : REEST MATING LIKELIHOODS isiNG
RETERMINING exits), Sha S Six NSR (;
Chili C -3 NEG-isoljiki-G RXELS ANDiOR
ECON f viii.2it signAS CSTANs). A close rive
^^3 - S - Sir i{{S^S.
:
. . xx w w w w xx x ... x was xx -x s a wa xx x . x as xx w, ..., x as xx w w x - xx w w x \ w, x w w w w w w ww w w w w w w w w w w x - w w w x t w w w w
S&N is gy Šiš, ri iF THE Sks:-83 : * {-g

& Strixx - X SS &r:33 - k3, £, RSS is:{3:
A
| | | | | || | | | | | | | | |
{ }{ } {
{
~~~~~~ |- - -
***;***,azaz**
|-4
****** * *
~· ~); -)b–†6uis epboluad

}|- Tiun
444
?~ -~?);
?- -}{uoqer o; :
Á?an
|
??a.a. a.
?;
~.- ~ )~;
~o, ;?
I?s||u|Voºdsºniulds/MoeSn ·
6unoel o
|
Cister -38
8 as : SY $8-
L.
‘. . .
5, 20A- 30-
rarr;
to --
s Processor'
H’L5E')Ð|N–TO(AuVZri)
U.S. Patent Aug. 19, 2014 Sheet 5 Of 9 US 8,809,787 B2
G. 33
| SA
Alaiogi sak
* band pass r st ississis
iter
respes
~~~~
:
•?~~
•
***
ø–ö-tjeõ?liq.-gwñ(inº5)·b?,.olza*-?|
&is,
Sigrai
&
FOs. A
Sigrai
(servative
Ex as a a
& S S S. &
I
sax
S. S. &
s
8.
s
Sh
SA
5000
Trans, H2O
(Band Model)
... &rs, is
8.0 i. duriuli
3 g is is : 8 88 g : 8 88 : 3 x 3.
Wavelength victors G. S3
wassass
.
8.
R
8
E:
g Target & Sur irradiance
8:
r
SriSr at 0.32-0.38 an 83rwiSr 5 S are
co& ea&s: coex &is ay &. &. ...a. &. R.. R. . is. is. a. is. .xi x}. is. a. -. .
x:X
E. : Es Kx is S. &c. S. & RS & RS is S. R. & 83 &g &S
r; p r in r is if r is in in is in in r r is in in
MS W: ms rS S3 & S. & S. & KS & & S & E3 E3 E3 & E3 &
Waveiength in r G. C
Muzzle
flash detector -
~~~~~~~
. .. . . . .. . . . . . Systein
r K- X: it
Acoustic
&s detector
US 8,809,787 B2
1. 2
GUNSHOT DETECTION SYSTEMAND Uncooled infrared sensors for an Integrated Sniper location
METHOD system were studied in 8. The system of 8 had a focal plan
array size of 320x240 pixels allowing a field of view of 20
RELATED APPLICATIONS (H)x15 (V) degrees with an accuracy of 2 mrad. The systems
weight was 5 pounds, frame rate 60 Hz, size 12.2 (L)x5.0
This application is a Continuation of International Appli (W)x4.1 (H) inches, noise equivalent input 5.6x10-12
cation No. PCT/IL2008/000105 filed on Jan. 23, 2008, which W/cm2, and power consumption 9V. The projected price of
is incorporated herein in its entirety. the device was about S10,000. The uncooled bolometric
detectors are typically significantly slower than the muzzle
FIELD OF THE INVENTION 10 flash, and thus the signal is Smeared over a longtime harming
the signal to noise ratio (SNR). These detectors are mostly
This invention relates to the field of detection and location sensitive to the 8-12 um range, where the signature is rela
of high speed photon emitting events, of firearms gunshots, tively low.
and particularly of muzzle flashes. Other counterSniper systems, such as relying on acoustic
15 signals (e.g. muzzle blast and bullet shock wave), may be
BACKGROUND OF THE INVENTION lighter and lower in cost than systems based on cooled detec
tors. However, the acoustic countersniper systems typically
There are various methods that can be used for snipers have low angular accuracy and performance which is reduced
detection and locating weapon fire, including Small arms. The in urban terrain, due to Sound reflections.
phenomena utilized by these methods include the so-called Also, detection of rifle, Sniper and Small arm shooting or
muzzle blast and flash; the shock wave, Vortex, and thermal firing can be in principle done by the solar blind UV (SBUV)
signature of the bullet in flight; and retro-reflection from the imaging technology and used for force protection and Snipers
Sniper's optical sight. Other phenomena, for instance distur detection. The UV signature of the firing is due to the sec
bances of the refractive index of atmosphere caused by Vor ondary burning of the residual gunpowder, ejected from the
tices shed by the bullet in flight, can potentially be utilized for 25 barrel. Nevertheless, this signature is also relatively small and
Snipers detection. may not provide usable detection range and acceptable false
One of the optical signals caused by weapon firing is the alarm rate. Some design and manufacturing of SBUV imag
muzzle flash, which is the incandescent flash at the weapons ing systems is done, for example, by Ofil LTD (http://www.s-
muzzle caused by the ignition of oxygen, the expulsion of buV.com).
burning powder grains and the expansion of powder gasses. 30 In principle, various detection methods can be combined
The phenomenon of muzzle flash is more clearly pronounced with each other. Also, a muzzle flash detected by any method
in various assault rifles, short barrel infantry weapons, and can be shown on a scenery image obtained by imaging with
“cut down weapons. For instance, in a short barrel, the bullet visible light. For example, technology of the Ofil LTD utilizes
can leave the barrel before the powder is completely burnt. In a bi-spectral visible-UV DayCor(R) camera and is aimed and
this case, the unburnt powder ignites in the air, giving off a 35 presenting Such combined images.
bright flash. For a shooter, muzzle flash presents a serious
problem: it increases the shooter visibility to the enemy and SUMMARY OF THE INVENTION
obscures the target view. As a result, the shooter using a
weapon generating muzzle flash must move quickly after There is a need in the art in facilitating detection and
firing to avoid return fire. 40 location of high speed photon emitting events, of firearms
Although muzzle flash can be partially hidden by flash gunshots, and of muzzle flashes by providing a novel fast
Suppressors or partially reduced by using cartridges with a event detection technique allowing effective detection. A pre
faster-burning gunpowder, so that the propellant gases will sented here novel technique, constructed by the inventors, has
already have begun to cool by the time they exit the barrel, this adaptations (versions and embodiments) useful for Such
is not always convenient for the shooter. For example, the size 45 detection.
of a device necessary to hide the muzzle flash from an enemy The main idea of the technique of the inventors applied for
can be too large. example for gunshot detection, is to utilize imaging of light in
For the side opposite to a shooter, a problem consists in a spectra of relatively short-living and low power muzzle flash
muzzle flash detection, which can be considered as one of components, which though allow achieving a relatively high
high speed imaging applications. For purposes of study, Such 50 useful detector signal. The technique may utilize one or more
detection can be done in the laboratory. of the following: filtering electromagnetic radiation for
At the present time, a few muzzle flash detecting systems acquiring Substantially a spectral range corresponding to rela
can be used in the battle field. Examples of such systems tively low light transmission in atmosphere; collecting and
include Radiance Technologies WeaponWatchTM, RAFA sensing electromagnetic radiation in near infrared (NIR) and/
EL’s SPOTLITE, Maryland Advanced Development Lab's 55 or shortwave infrared (SWIR) ranges; acquiring pixel images
VIPER. The VIPER equipment, for example, consists of a with a relatively small integration (exposure) time or high
mid-wave infrared (MWIR) camera, together with real-time imaging frame rate; acquiring sequential images with a rela
signal processing, magnetic compass, and user display and tively small dead time between them; acquiring images with
alarm. It is advertised as providing gun detection within 70 a relatively wide field of view (FOV) for a pixel or for a given
msec after gunfire and geolocation of the firing event. Using 60 number of pixels at the light detector; acquiring multipixel
MWIR-camera also allows concurrently performing forward images of a scenery; processing outputs of the photodetector
looking infrared imaging (FLIR) of a region of interest. (imager) pixels for detection of a flash-type signal portion in
The known snipers detection methods also suffer from the detected light. For example, the processing may select, for
various problems. One of the problems associated with any pixel, a signal portion indicative of muzzle flash type
MWIR cameras, is that these cameras are expensive and 65 intensity variation in time. For another example, the process
bulky. Most of them are based on cryogenically cooled fast ing may estimate, for any pixel, a likelihood of occurrence of
refresh-rate detectors. a muzzle flash event, e.g. by comparing a signal obtained
US 8,809,787 B2
3 4
from the pixel with the time signature of muzzle flash. The off. In some embodiments, an imager or imager arrangement
processing may be organized in layers (stages), consecutively of the technique of the inventors has a long-wavelength cutoff
processing a Smaller number of pixels. For example, the lay longer than 3 um (i.e. the SWIR region edge) while this
ers may be organized to consecutively reestimate the likeli imager is used for NIR/SWIR imaging; such a long-wave
hoods of occurrence of muzzle flash events for a smaller length cutoff can be useful for example when the imaging is
number of pixels at each layer. Processing may be parallel at to be done in a broad region within the NIR/SWIR range. In
Some layer; however the technique of the inventors may as Some other embodiments the imager has a cutoffshorter than
well utilize non-parallel layers of processing for reducing 3 Lum (this relates to those qualities of the typical shorter
databandwidth. For example, the processing may have a layer wavelength imagers that have been discussed above). In some
at which a signal obtained from a pixel is compared to signals 10 embodiments the cutoff is between 0.75 and 1 lum, in others it
obtained from adjacent or close pixels, and if time signatures is 1 um and 1.4 Lum, in Some others it is between 1.4 um and
of a group of such pixels are similar, a possible muzzle flash 1.65um, in yet others it is between 1.65um and 1.8 Lum, in yet
alarm is Suppressed. The processing layers may be connected Some others it is between 1.8Lum and 2.5um. In some embodi
by pixel selection units, decreasing number of “candidate' ments, it is the sensitivity maximum that is at a wavelength
pixels processed at the later stage. Each pixel selection unit 15 shorter than 3 um or alternatively constrained within the
transmits a list of the “candidate’ pixels from an earlier layer specified ranges. Also, in Some adaptations, the imager has a
to a later layer. For example, a pixel selection layer may be sensitivity maximum being at a wavelength longer than 3 um,
adapted to transmit a limited or a constant number of 'can so as to utilize a relatively flat region of growth of the sensi
didate’ pixels from each frame, by selecting pixels with the tivity band, but has a cutoffshorter than about 5um, so as to
highest likelihoods of occurrence of muzzle flash. If output of utilize this region of growth not too far away from the band's
a processing layer is not the likelihoods, but for example a maximum, i.e. while the sensitivity is already relatively high.
signal portion indicative of muzzle flash type intensity varia The selection of the photodetector material and sensitivity
tion in time (as mentioned above), the consecutive pixel band parameters can depend for example on type of event or
selection layer generates pixels likelihoods. In this sense, a muzzle flash to be detected and on the presence/absence of
pixel selection layer is complementary to its preceding pro 25 clutter and the low light atmosphere transmission optical
cessing layer. However, the inventors do not require always filter. In particular, in some embodiments the wavelength of
having a pixel selection layer for each processing layer: out the sensitivity band maximum is close to or is aligned with
puts of the latter may be for example directed to a memory one of the atmospheric absorption peaks, so that the imager is
rather than to the former. In addition to the list of the “candi utilized efficiently. The imager operation is efficient if it
date pixels, a processing layer will need additional data for 30 occurs in a range where the imagers sensitivity is more than
(re)estimating muzzle flash likelihoods. For example, tempo 20% (or, according to alternative definitions, 35%, 50%, and
ral and/or spatial information on outputs of various process 70%) than the imager's sensitivity maximum. Thus defined
ing stages and/or of the Photo Detector Array (PDA, e.g. range (i.e. defined according to one of the alternatives) con
photo diode array) may be needed. The needed data may be stitutes a predominant range of the imager sensitivity. How
stored in a memory unit configured for access by the layer. 35 ever, it should be noted, that the efficient use of imager is not
Before data are stored, they may be compressed or sampled a requirement: the sensitivity of the PDA may be sacrificed
with a reduced sampling rate (relatively to the output of the for other factors, such as response time. Also, a not so effi
corresponding data source stage). The compression can be ciently used imager of one type may still be more sensitive
performed in a compressing module. The processing may than an efficiently used imager of another type. Any imager
have analog and digital layers, where the first of digital layers 40 can be used as long as it is suitably sensitive.
is preceded by an analog-to-digital converter (ADC). Additionally, for NIR/SWIR imaging, the sensitivity maxi
The technique of the inventors can be used for detection mum of the imager may be kept at a wavelength longer than
and location of strobe light sources, pulsed lasers, lightnings, 0.75um for achieving higher efficiency. Although some NIR/
as well as antitank missile launches and shell firings. SWIR imaging may be performed if the sensitivity maximum
With regards to the wavelengths useful for detection of 45 is shorter than 0.75 um and the cutoff is longer than 0.75 um.
muzzle flash, the inventors have considered that the detection It should be understood that the above sensitivity consid
can be facilitated in some of its aspects by using imagers (e.g. erations are relevant for the sensitivity as measured in circum
based on PDAs), predominantly sensitive in near-infrared stances of the imager operation. The sensitivity is dependent
(NIR) and/or short-wave infrared (SWIR) region: in contrast on temperature. The operation temperature of the PDA may
to MWIR/LWIR images, NIR/SWIR imagers typically need 50 be lower than -30° C., or between -30° C. and 0° C., or
not be cooled. Partially due to this fact, partially due to other between 0° C. and a PDA’s ambient temperature (while it is
reasons, NIR/SWIR imagers also may be lighter, less expen higher than 0° C.), or higher than the PDA’s ambient tem
sive, and less power-consuming than MWIR/LWIR imagers, perature. The latter option for the PDA selection (i.e. for the
especially when the MWIR/LWIR imagers are considered PDA operation temperature selection) is used in some of the
together with their respective coolers and portable power 55 adaptations in which the portability is a goal; it would require
supplies, NIR/SWIR imagers (particularly, NIR/SWIR no cooler or would require only a weak cooler. The ambient
PDAs) include many types of silicon imagers, which typically temperature is the temperature which the PDA would have
are sensitive up to about 1050 nm, and many types of InCaAs had the device be not turned on. If there is anything external
imagers (i.e. PDAs), which typically are predominantly sen to the device (e.g. Support) or anyone else (e.g. the user) that
sitive in a band starting from around 950 nm and ending 60 heats the imager, this would affect the ambient temperature,
somewhere between 1700 and 2500 nm (NIR/SWIR), which by default would be a weather temperature. For the
depending on the specific InGaAs composition. SWIR imag operation temperature to be in the desired range, cooling,
ers include some types of Mercury Cadmium Telluride when necessary, may be provided by cryogenic coolers in the
(MCT) imagers. Imagers' sensitivity bands depend not only coolest of these ranges or, for example, by thermoelectric
on their materials, but also on their structures (e.g. quantum 65 coolers in hotter ranges. Cooling may be or may not be
structures). A typical sensitivity band has a sharp decline after necessary in the hottest of the ranges. For the above tempera
a sensitivity maximum and ends with a long-wavelength cut ture ranging it was considered that weather/ambient tempera
US 8,809,787 B2
5 6
ture is often above 0°C.; it is understood, however, that the The technique of the inventors may utilize acquiring
device may be used at the ambient temperature which is sequential images with a relatively small dead time between
below 0°C. and, possibly, even -30°C.: in this case the device them. The dead time can be define, as a time between reading
may be configured or adapted to have not to have a cooler or a pixel and Subsequent resetting a pixel. The signal (e.g.
have only a weak cooler. If cooling is used, then focusing charge) generated at the pixel during the dead time gets lost.
optics (and/or filter) may also be mounted within the respec In Some adaptations of the technique the dead time is selected
tive cooled region. to be less than the detected muzzle flash duration (or a pre
For the purposes of the present application, the limits of determined part of the detected muzzle flash duration, such as
infrared ranges are defined here: NIR 0.75-1.0 um; SWIR one tenth). In particular, the technique may be adapted to use
1.0-3 um; MWIR 3–8 um; LWIR 8-14 lum. Thus defined 10 dead time under a millisecond. The desired dead time can be
ranges correspond to commonly accepted infrared nomencla set by an appropriate shutter scheme, for example by a rolling
tures. For example, within the SWIR range there are several shutter.
absorption peaks of water and carbon dioxide: at approxi The technique of the inventors allows detection of muzzle
mately 1.15um, 1.39 um, 1.9 um, and 2.7 um. The absorption flashes in a relatively wide PDA field of view or per pixel field
peaks are commonly known and some of them are shown on 15 of view. Consider, for example, a muzzle flash occurring at a
some of the figures referenced below. certain distance that gets projected on a pixel. The same
The technique of the inventors may utilize a detector unit muzzle flash occurring further from the imager will then be
configured to have, inter alia, a narrow spectral band filter projected on a spot Smaller thana pixel; and the pixel then will
corresponding to low light transmission in the atmosphere. generally collect more clutter and less signal (in particular
Such a filter may increase, in some scenarios, signal to clutter because a smaller portion of signal will propagate towards the
and signal to shot noise ratios. In typical muzzle flash detec imager and because this Smaller portion will undergo more
tion systems, the spectral band utilized for detection is wide absorption in the atmosphere). This will result in a lower
and corresponds to the highest transmission atmospheric win signal to clutter ratio and a lower chance of successful muzzle
dows. In some embodiments of the invention, detection may flash detection. Thus, for a desired distance of detection a
be performed in the narrow range(s) of the least atmospheric 25 pixel should not collect light from a too broad region. Though,
transmission, because though a large portion of the muzzle it can be concluded that the projection spot for a muzzle flash
flash signal would be lost with the introduction of the narrow may be smaller thanapixel (in one or two dimensions) and yet
filter, the reduction of sunlight clutter and the shot noise allow detection, if optical resolution allows.
would be more drastic and would overcompensate the reduc It should be noted, that the useful number of pixels is tied
tion in the informative optical signal, if only the remaining 30 with various factors such as the photodetector architecture
portion of the latter is above the sensitivity threshold. and the complexity of processing used for detection. Thus, the
The technique of the inventors may utilize acquiring useful field of view of an exemplary PDA's pixel and the field
images with a relatively small integration (exposure) time of view of the photodetector itself depend on a type of detec
and/or high frame rate. In particular, in some adaptations of tion being implemented. The field of view of a single pixel
the technique, the integration time is selected to be substan 35 may be selected to be relatively wide, so as to correspond, for
tially equal or smaller than the duration of existence of the example, to the size of a muzzle flash occurring at a distance
muzzle flash components which to be detected (this duration smaller than a few tens of meters. Typically, muzzle flash
can be defined as a time interval between the moments at would occur at a distance larger than that, and would cover
which the detected radiation intensity equals half of the maxi area Smaller than a pixel. In accordance with the above, when
mum detected radiation intensity). The integration time can 40 muzzle flash occurs, the pixel signal would be a sum of a
be defined as a time between resetting a pixel and Subsequent signal due to the background and of a signal due to the muzzle
reading a pixel or a duration of a time period during which flash. The background typically changes slowly, so if a fast
photons collected by a pixel are transformed to electrons of a change in the total pixel signal is detected, this makes the
single signal reading from a pixel. The rationale behind the pixela candidate for the processing aimed at checking muzzle
choice of the integration time is in that imaging for a time 45 flash occurrence. The wider the pixel’s field of view, the
longer than the duration of the detection muzzle flash portion smaller the relative part of the muzzle flash signal in the total
would collect rather clutter and noise thana useful signal. The pixel signal, for the desired range of detection. Therefore,
technique may use integration time between The PDA inte there is a trade-off between the achievable range and pixel
gration time may be between 10-2 s and 5.0-10-3 s, or field of view. While the desired range is achieved, the total
between 5.0-10-3 s and 2.0-10-3 s, or between 2.0-10-3 s and 50 field of view can be increased by increasing a number of
5.0-10-4s, or between 5.0-10-4s and 10-4s, or it might be less pixels in the detector. The increased total field of view is
than 10-4s. preferred when it is desired to detect shots from as many
The desired integration time can be set by an appropriate directions as possible. However, using a too large pixel count
shutter, particularly an electronic shutter, it should be noted might require too expensive optical and computational hard
that the technique of the inventors may utilize acquiring 55 ware, and too much power for operation. In some adaptations
images with a relatively long integration time, for example of the technique of the inventors the divergence angle of the
longer than the pulse duration. Nevertheless, in the case of field of view of a single pixel is selected to be between 0.1 and
integration time longer than the pulse duration, the Smaller is 0.2 degrees, or between 0.2 and 0.5 degrees, or larger than 0.5
the integration time, the better is the signal to noise ratio and degrees. In some adaptations of the technique the field of view
signal to clutter ratio. Moreover, better signal to noise and 60 of the photodetector is selected to be between 30 and 50
signal to clutter margins allows enlarging the Instantaneous degrees, or between 50 and 90 degrees, or larger than 90
Field Of View (IFOV) (e.g. by decreasing the focal length of degrees, in at least one dimension. The desired field of view
the imaging lens) and thus, for a given pixel count, allows may be created by appropriate optics with a focal length
increasing the entire Field Of View (FOV). With integration selected to project the field of view on the detector surface.
time shorter than the pulse duration, the inventors' technique 65 The technique of the inventors may advantageously utilize
can also benefit from analyzing the time signature of the acquiring multipixel images of Scenery. In some adaptations
detected signal. of the technique the imager is selected to have between
US 8,809,787 B2
7 8
10,000 and 100,000 pixels, or between 100,000 and 1,000, The staged (layered) processing allows decreasing power
000 pixels, or it may be suitably more than 1,000,000 pixels. consumption of the detector and weight of the needed power
Using the multipixel imaging, muzzle flash locating may be Supply and increases a maximum allowed input data band
performed relatively accurately and precisely. The muzzle width (e.g. number of pixels in the PDA and the breadth of the
flash lateral location is, to a large degree, determined by the field of view) and the portability of the detector.
imaged direction of a pixel or group of pixels which detected Layer(s) of processing may be combined by Such pixel
the muzzle flash. The muzzle flash longitudinal location is, to selection utilities, as Constant False Alarm Rate module(s)
a large degree, determined by the intensity of the received (CFAR modules), which confirm/suppress the candidate
signal and shape of the pixel group which detected the muzzle alarms for the same number of suspicious pixels perframe (or
flash. 10 a certain number of pixels of the PDA) or time interval. The
The muzzle flash detecting processing, which particular use of CFAR technique allows not to jam a net processing
example is parallel processing, may be performed by a unit stage of the detection system, but to keep maximal probability
configured to apply Such operations to the pixels outputs as of muzzle flash detection. The CFAR module is one of pos
single or multiple (in particular double) signal differentiating, sible pixel selection modules.
15 The parallel processing can include parallel analogin-pixel
peak detecting, band-pass filtering, match filtering and/or signal processing done by a corresponding processing (sub)
other operations aimed at Selecting or detecting muzzle flash unit. The parallel analog processing unit can be based, for
type signal(s) from the detected signals. An output signal example, on the hybrid detector technology not requiring
obtained as a result of a single pixel processing (e.g. an output compromising in-pixel processing power or array fill factor.
signal of obtained from a pixel of the parallel processing unit) In some embodiments the fill factor of the PDA is between
may be interpreted for estimating likelihood that the corre 60% and 75%, in some other it is between 75% and 90%; in
sponding PDA pixel has detected a muzzle flash. For some other embodiments it is higher than 90%. The fill factor
example, the likelihood may be contained in the output of the may be also outside of the specified regions, however, typi
match filter. If the likelihood is larger than a certain threshold, cally, the higher the fill factor, the higher the chance of muzzle
the event may be interpreted as a muzzle flash or a candidate 25 flash detection for remote muzzle flashes. In particular, close
muzzle flash, and a signal (alarm) intended to inform an to 100% fill factor almost eliminates the chance that the
operator or a user about the shot may be produced, or a remote muzzle flash will be projected on the dead area (where
candidate alarm may be transferred to a next layer or stage of the PDA is not sensitive). In addition, even for not so remote
processing. muzzle flashes, the higher is the fill factor, the higher is the
As it has been mentioned above, the processing unit can be 30 collectable signal. The hybrid detector can be fabricated as a
organized in layers or stages. The filtering performed by one Read Out Integrated Circuit (ROIC) with in-pixel signal pro
stage of the muzzle flash selecting processing unit can effec cessing flip-chip bonded to photodiode or other photodetector
tively reduce an amount of data for processing to be per array.
formed at further stages, thus allowing to apply to the reduced The parallel in-pixel signal processing may also be done in
amount of data a more complex processing and to reduce 35 a separate electronics module (not in the ROIC). The parallel
requirements for the technical characteristics needed at a processing of the PDA pixels outputs can include analog-to
particular stage of the muzzle flash detector. In some embodi digital conversion and processing of the digital data by a field
ments of the technique of the inventors, the processing is programmable gate array (FPGA). The term parallel should
performed in stages, with a parallel stage aimed at analyzing be distinguished from simultaneous in the context of the
time-dependences of pixels outputs being the first or one of 40 present application: the parallel processing can be either
the first stages. The parallel layer of processing can be com simultaneous or not. If pixels outputs are processed indepen
bined with a module for selection of pixels which most likely dently of each other, such processing is parallel, indepen
detected muzzle flash. A further (e.g. the second) stage thus dently on order in which their processing is performed. The
can receive a list of candidate pixels from the module. This parallel processing is parallel to an extent allowed by cross
further layer of processing then may process this pixel list, by 45 talk.
obtaining from a corresponding memory unit data needed for After the first, time-dependence analyzing and possibly
reestimation of likelihoods of muzzle flash event for pixels parallel, layer or stage of processing, there may be other
from the list. The data needed for reestimation may include layers of processing, further testing pixels for presence of
historical data of pixels from the list and historical data of gunshot events. For example, a second (or further) layer of
pixels situated close to the pixels from the list. The historical 50 processing may be configured to operate in the following way.
data may include previously generated outputs of various It may select and analyze one or several candidate or Suspi
processing stages, pixel selection modules, and of the PDA. cious events, i.e. signals coming from pixels which have been
The historical data, before they are stored in the memory, may determined by the first layer processing, for example includ
be compressed. For example, PDA outputs may be divided so ing a CFAR module, to have high likelihood of detecting
that one replica will follow to the first layer of processing and 55 muzzle flash events. For each of the Suspicious pixels (can
another replica(s) will follow to a compression module(s)and didate pixels) the second layer of processing reestimates the
then to a memory unit(s). Thus, the memory unit can be used likelihood that this pixel detected a muzzle flash. Since the
to store historical data for all pixels. second layer analyzes a smaller dataset that the first layer, it
Likewise, the data to be used for reestimation may be or may use more Sophisticated algorithms or processing, requir
include data obtained after a candidate event, in addition to 60 ing higher computing time and power per pixel, to better
the historical data, as the processing does not have to be distinguish between muzzle flash, noise and clutter. For this
immediate, and in fact may benefit from taking into account purpose, the second layer of processing may use signals
information gathered after the candidate event had occurred. obtained from neighboring or close pixels and/or signals
In this case, to obtain the same quality of detection, the obtained at close time moments. For example, the second
memory unit may be used to store only a portion of volume of 65 layer may be configured to check whether an eventis repeated
the historical data, because the data may be gathered more for several adjacent or close pixels and propagates along a
intelligently when the candidate pixels are known. straight line on the PDA. If the case is such, the source of the
US 8,809,787 B2
9 10
moving event (i.e. of a sequence of events) may be identified which is 7 additional orders of magnitude. In some preferred
as a moving object rather than a muzzle flash. Thus the embodiments the decrease in the number of candidate events
invented detection system eliminates events lasting too long due to the first, in particular parallel, layer is smaller than the
time or passing too long distance: for example, level-1 may decrease in the number of candidate events due to a further
eliminate most of the events lasting too long within one pixel layer (e.g. the second layer).
over time, while level-2 may eliminate events lasting too long In this connection it should be understood, that the layered
in the entire frame, though lasting for a relatively short time in architecture can facilitate detection of very short events, such
each of the involved pixels. The closeness or relative location as muzzle flashes, while it the technique of the inventors is
in space of the involved pixels allows inferring that the aimed at operation with a high rate (decision-making rate)
sequence of events is not a sequence of the independent 10 and an efficient power use scheme.
events, but a single moving event. The technique of the inventors may combine detection in
The second layer of processing may be fed with data in the different spectrums, such as SBUV or visible, by using one or
Constant False Alarm Rate regime and confirm/suppress the more PDAs sensitive to one or more wavelengths. For
candidate events/alarms for a constant predetermined number example, the detection of the flash intensities corresponding
of Suspicious pixels without jamming the detection system. 15 to different wavelengths can be done with one detector using
The confirmed alarms can be passed to an operator or a user time and/or pixel multiplexing. In particular, a secondary
of the detection system (e.g. by an optical signal or a sound). visible light or NIR imaging can serve for independent detect
The reasons why some of the preferred embodiments of the ing of muzzle flashes and for reducing the false alarm rate
invented technique use at least two layers of signal processing (FAR) of a prime (in this example SWIR) imaging by con
can be illustrated by the following consideration. The system firming or Suppressing Suspicious events. In other words, the
of the invention may need to make for example of decisions results of the prime imaging may be verified by the results of
per second. This number is obtained assuming a frame rate of Secondary imaging.
500 per second and a PDA of 200,000 pixels. The inventors Also, visible light or NIR imaging can serve for deduction
assumed that a final false alarm rate (FAR) of not less than of the background by Subtraction of images, because, for
approximately /3 hours is desired. Then, only one false alarm 25 example, in daylight the muzzle flash is stronger in the SWIR
is allowed in every decisions. In Such a case, if one would range, while the background is stronger in NIR or Visible
apply the most Sophisticated algorithm or processing to each light.
of the pixels all the time (at the PDA framerate), the algorithm The muzzle flash detector may be incorporated into a gun
would use a lot of computing power. Thus, to save the power, shot detection system also including an acoustic or any other
the inventors in some cases choose to split the processing into 30 gunshot detector. Such a system can perform concurrent
layers, to utilize more efficient computing power Scheme. detection of the muzzle flash and of some other (e.g. acoustic
In some of the embodiments the reduction of data band or SBUV) signal associated with the gunshot, e.g. of muzzle
width is 3-9 orders of magnitude per layer. The reduction of blast and/or bullet shock wave. This double (e.g. optic/acous
databandwidth can be understood as a reduction of number of tic) detection scheme is aimed at decreasing the FAR and
potential alarms: the layers reduce their number from the 35 increasing the ratio of probability of gunshot detection (PD)
original pixel rate (number of pixels in PDA times the sam to the FAR. The user is notified about the gunshot only if both
pling rate) to the maximum allowed FAR. The layers may optical detector and the acoustic detector identify a gunshot
eliminate also true events, however the probability of detec (i.e. if the optical detector “suspicion” or alarm is confirmed
tion (Pd) is kept high. For example, detection with a probabil by the acoustic detector within a short period of time needed
ity of success of over 80% and allowed FAR of not more than 40 for sound to cover the distance to the user).
1 per 3 hours (i.e. approximately 1 per 104 seconds) may be In some of the preferred embodiments the inventors’ sys
considered. By utilizing a PDA with 2x105 pixels and frame tem includes or is associated with a control unit configured for
rate of 500 per second, a designed detection system obtained at least one of the following: for processing the output signals
2x105x500-108 samples per second. Accordingly, the speci of the parallel processing unit for confirming/Suppressing the
fied FAR corresponded to 1 false alarm per 1012 sampled 45 possible alarms by constructing information contained in spa
pixels. The first layer detected at least 90% of true events, and tial and/or temporal features of the detected optical signal(s):
falsely detected muzzle flash in average in 1 case per 105 for processing signals obtained from multiple detectors (e.g.
sampled pixels (i.e. defined some pixel as a candidate while optical, such as PDA, or acoustic, or two optical for different
that pixel did not correspond to the muzzle flash). The wavelength ranges) for confirming/Suppressing the possible
designed second layer detected at least 90% of true events, 50 alarms and/or for determining a distance to the flash; for
and falsely detected muzzle flash in average in 1 case per 107 defining the PDA working parameters and/or parallel pro
sampled pixels. The combined performance of the system cessing parameters based on external conditions (e.g.
therefore was characterized by the probability of detection of weather, lighting); for determining a weapon used by the
81% (90% times 90%) and the FAR of 10-4 S-1 (10-5x10-7x shooter (e.g. by comparing the detected data with reference
(2x105)x500s-1). 55 indicative of different muzzle flashes signatures, or by mea
In accordance with the above, in Some embodiments, the Suring the time between muzzle flashes in case of bursting
second layer receives data from the first layer through a CFAR fire). In accordance with the above, the control unit may be
utility, structurally included either into the first layer or the configured to implement the second and/or further layers of
second layer and selecting a constant number of the most processing. Thus, the control unit facilitates the detection of a
Suspicious events for processing in the rest of the second 60 muzzle flash event.
layer. Selecting more suspicious events than the second layer The technique of the inventors may be aimed at a high
can handle would create an overflow, e.g. a fail of a cycle. capability of use by a human having a restricted immediate
Selecting too few events would increase the FAR. Operating access to various carrying equipments, for example a soldier
in its normal mode, the second layer further reduces the FAR, or a policeman. To this end, weight of Some of the embodi
to the maximal allowed level. For example, the first layer may 65 ments may be selected to be between 1 kg to 3 kg, or between
reduce the data rate from the pixel rate of about/sec to about/ 0.3 to 1 kg, or lighter than 0.3 kg. Such selection is facilitated,
sec and the second layer may reduce the FAR to a desired/sec. when the PDA does not require a cooler, in particular a cryo
US 8,809,787 B2
11 12
genic cooler (e.g. Stirling cooler), and a relatively heavy According to a broad aspect of the invention, there is pro
portable power source for feeding the cooler (consequently, vided a method for use in detection of a muzzle flash event.
the system may need only a relatively portable light power The method may include one or more of the following:
source). This relates to the above references to the range of the (a) focusing on a Photo Detector Array (PDA) electromag
PDA working temperatures. In fact, some of the inventors netic radiation, being at least partially within the near infrared
technique's adaptations are configured to employ a working (NIR) and short wave infrared (SWIR) spectrum;
temperature as of natural environment (or slightly larger, due (b) focusing on a pixel of a PDA electromagnetic radiation
to heat dissipation in the device). However, it should be from a relatively large pixel field of view (FOV);
understood, that the high portability is not a requirement; the (c) acquiring multipixel images of a scenery by a PDA,
system may be heavy, especially when it becomes necessary 10 comprising a relatively large number of pixels;
due to other reasons, for example when the system needs to be (d) focusing on a PDA electromagnetic radiation from a
shielded and/or when the system needs to stay operative for relatively large field of view:
long periods of time and thus requires a highly capacitive (e) filtering electromagnetic radiation so as to allow sens
ing by a PDA of Substantially a spectral range corresponding
power source. For another example, the system may be sta 15 to relatively low light transmission in atmosphere;
tionary mounted for constantly observing a desired scenery, (f) sensing by a PDA electromagnetic radiation, being at
which may be a Subway station, or a street with governmental least partially within the near infrared (NIR) and the short
offices, or any place where a terrorist or bandit attack might be wave infrared (SWIR) spectrum;
expected, and therefore it may not need portability. Likewise, (g) using a relatively small integration time for sensing
the inventors system may be mounted on a tank, or a car, or electromagnetic radiation by a PDA:
a bus, etc. The system may be fed with a constant power (h) using a relatively small dead time for sensing electro
Supply, similarly to a home lamp or a desktop computer, and magnetic radiation by a PDA:
it may be configured to utilize a MWIR/LWIR imager and a (i) multiplexing pixel signals originating from a PDA into
cooler and benefit from a stronger muzzle flash MWIR signal. at least two replicas;
However, the inventors have realized that the latter MWIR/ 25 () recording a replica of pixel signals originating from a
LWIR imager and a cooler are not necessary, and do not have PDA into a memory;
to be utilized when the relatively high portability is desired. (k) applying a processing to pixel signals originating from
Contemplating on the latter case, the inventors have found a PDA, the processing being adapted for use in selection of
that the portability may be increased, when the system of the candidate pixels, whose signals are substantially similar to
invention is more specifically configured for a specific mis 30 the temporal-spatial signature (i.e. temporal and/or spatial) of
Sion. The inventors have considered that for Some missions, muzzle flash;
the mission duration is typically between 6-12 hours, while (l) utilizing a relatively portable system for the detection;
for some other missions the duration may be between 1 and 6 (m) utilizing a relatively low capacity portable power Sup
hours, or between 12 and 24 hours, or longer than 24 hours. In ply for the detection;
fact, the expected mission duration depends on a scenario, 35 (n) using a PDA at a temperature equal to or higher than a
which caused the need to use the inventors’ system, on the temperature of natural environment for the detection;
scenery, the length of the day in the season, the capabilities of (o) not utilizing a cooler for a PDA being used for the
the systems user to stay in engagement or to replace or detection;
recharge the portable power Supply. Thus, as the inventors (p) focusing on a Photo Detector Array (PDA) electromag
have found, the capacity of the portable power Supply may be 40 netic radiation wherein the focusing is optimized for an object
selected so as to provide the system a desired operative dura distance less than 50 meters;
tion, for example in one of the ranges of the mission duration (q) using a Photo Detector Array (PDA) with a fill factor
above. The power Supply, if it is light, may provide enough larger than 60%.
electricity even for a larger number of hours. In particular, the According to a broad aspect of the invention, the process
capacity of the portable power Supply (e.g. an accumulator or 45 ing for the detection of a muzzle flash event may include at
a battery) may be less than 0.1 W, or between 0.1-1 W and least one of the following:
between 1-10 W or larger than 10W. It should be understood, (a) eliminating spatial background portion from the pixel
that modern muZZle flash detecting devices are typically signals:
equipped with power supplies of more than a 50W capacity (b) eliminating relatively slowly changing portions from
and therefore have a reduced portability, if at all. 50 the pixel signals;
Additionally, it should be noted, that the allowed weight of (c) selecting pixel signals’ portions Substantially similar to
the system may be a decreasing function of the expected the temporal signature of muzzle flash;
mission duration, because the user may need to carry addi (d) and analog processing of time dependence of the pixel
tional supplies (e.g. food, water, bullets) with him or her to the signals, the processing being adapted to generate Substan
mission, and the total weight may be limited. Therefore, for 55 tially an estimate of likelihood that a pixel detected muzzle
Some embodiments the inventors select the capacity of the flash;
power Supply and the system weight not independently, but in (e) a parallel processing of time dependence of the pixel
combination. This way, a relatively high portability and a signals, the processing being adapted to generate Substan
relatively high usability of the device of the inventors can be tially an estimate of likelihood that a pixel detected muzzle
achieved. 60 flash;
In this connection, it is reiterated, that in accordance with (f) a digital processing of a time dependence of the pixel
the features of the invention herein presented, the technique signals, the processing being adapted to generate Substan
of the inventors can be effectively utilized for detection not tially an estimate of likelihood that a pixel detected muzzle
only of muzzle flashes, but also for detection of other flashes flash;
and short events, for example of Strobe light sources, pulses 65 (g) selecting the candidate pixels by comparing a Substan
lasers, lightnings, antitank missile launchers, and shell fir tial estimate of likelihood that a pixel detected muzzle flash
1ngS. with a threshold common for a plurality of operating pixels;
US 8,809,787 B2
13 14
(h) selecting the candidate pixels by comparing an estimate The PDA dead time may be shorter than 1 millisecond.
of likelihood that a pixel detected muzzle flash with a thresh The replicated pixel signals may be compressed before the
old obtained using replicas of the pixel signals; recording.
(i) selecting Substantially the same number of the candidate The processing for the detection of a muzzle flash, while
pixels from Subsequent frames; 5 including eliminating relatively slowly changing portions
() at least two processing stages, a later of the stages from the pixel signals, may include at least one differencing
applying a processing for the selection of the candidate pixels the pixel signals. In particular it may include the second order
to a smaller number of the pixels than an earlier of the stages; differencing of the pixel signals.
(k) at least two processing stages, a later of the stages The processing may include selecting pixel signals’ por
applying, to the pixels, a processing for the selection of the 10 tions Substantially similar to the temporal signature of muzzle
candidate pixels, using more processing time per pixel than flash. Such portions for selection may be pulses in the form of
an earlier of the stages; a peak. The peak to be selected may be of a predetermined
(1) selecting, as the candidate pixels, those of the pixels, duration.
whose signals together with their vicinity pixels signals The analog processing adapted to generate substantially an
present Substantially the spatial signature of muzzle flash; 15 estimate of likelihood that a pixel detected muzzle flash may
(m) eliminating from the selection those of the pixels, include at least one differencing of the pixel signals. In par
whose signals together with their vicinity pixels signals ticular, it may include the second order differencing of the
present Substantially the temporal-spatial signature of a mov pixel signals. It may include passing the pixels signals
ing light source: through a matched filter, adapted to match the temporal sig
(n) confirming, as the candidate pixels, those of the pixels, 20 nature of muzzle flash. The substantial estimate may be a
whose signals together with their vicinity pixels signals, Voltage potential at an output of an analog circuit performing
obtained after their initial selection of the candidate pixels, the analog processing.
increase an estimate of likelihood that a pixel detected muzzle The selecting candidate pixels by comparing an estimate of
flash; likelihood that a pixel detected muzzle flash with a threshold
(o) eliminating from the selection those of the pixels, 25 may use common threshold within at least two groups of
whose signals are not accompanied by an acoustic signal with adjacent pixels, while threshold may be varying between the
a signature Substantially similar to the signature of a shot. groups.
The electromagnetic radiation being focused on the Photo According to a broad aspect of the invention, there is pro
Detector Array (PDA) may in part or in whole be within the vided a device foruse in detection of a muzzle flashevent. The
NIR spectrum. In particular, it can be fully within the NIR 30 device may include a PDA and one of the following features:
spectrum. It can also be at least partially within the SWIR (a) optics adapted to focus on the PDA electromagnetic
spectrum. In particular, it can be fully within the SWIR spec radiation being at least partially within the near infrared
trum. (NIR) and short wave infrared (SWIR) spectrum;
In some embodiments, the focusing is optimized for an (b) optics adapted to focus on a pixel of the PDA electro
object distance larger than 50 meters (i.e. the distance from 35 magnetic radiation from a relatively large pixel field of view
the muzzle flash to the detection device). In some embodi (FOV);
ments, the focusing is optimized for an object distance less (c) optics adapted to focus on the PDA electromagnetic
than 50 meters. radiation from a relatively large field of view:
The relatively large pixel field of view (FOV) may be (d) a filter of electromagnetic radiation accommodated so
between 0.1 and 0.2 degrees, or between 0.2 and 0.5 degrees, 40 as to allow sensing by said PDA of substantially a spectral
or it may be larger than 0.5 degrees. range corresponding to relatively low light transmission in
The device's PDA may include a relatively large number of atmosphere;
pixels. The relatively large number of pixels may be between (e) a shutter, allowing to use a relatively small integration
10,000 and 100,000 pixels, or between 100,000 and 1,000, time for sensing electromagnetic radiation by the PDA:
000 pixels, or it may be more than 1,000,000 pixels. 45 (f) a shutter controller, adapted to define a relatively small
The relatively large field of view may be between 30 and 50 dead time for sensing electromagnetic radiation by the PDA:
degrees, or between 50 and 90 degrees, or it may be larger (g) a multiplexer, adapted to divide pixel signals originat
than 90 degrees. ing from the PDA into at least two replicas;
The filtering may be performed in the spectral range at least (h) a memory, storing a replica of pixel signals originating
partially including the trough of low atmospheric light trans- 50 from the PDA:
mission situated around 1.15 um. Additionally or alterna (i) a processing unit adapted to process pixel signals origi
tively, it may be performed in the spectral range at least nating from the PDA, the unit being adapted for use in selec
partially including the trough of low atmospheric light trans tion of candidate pixels, whose signals are substantially simi
mission situated around 1.39 um; the trough of low atmo lar to the temporal-spatial (i.e. temporal and/or spatial)
spheric light transmission situated around 1.9 um; the trough 55 signature of muzzle flash;
oflow atmospheric light transmission situated around 2.7Lum. () a portable power Supply, configured to provide the
The sensed by the PDA electromagnetic radiation may be device with electricity for a relatively short time of the device
at least partially within the NIR spectrum. In particular it may operation.
be fully within the NIR spectrum. It may be at least partially The device's shutter may be an electronic shutter. The
within the SWIR spectrum. It may be fully within the SWIR 60 shutter controller may utilize a rolling shutter scheme for
spectrum. operating the shutter.
The PDA may be a CMOS PDA. Also, it may be an intra The device's PDA may be sensitive within a part or the
cavity PDA. whole of the NIR and SWIR spectrum, or of the NIR spec
The PDA integration time may be between 10-2 s and trum, or of the SWIR spectrum.
5.0-10-3 s, or between 5.0-10-3 s and 2.0-10-3 s, or between 65 The device's relatively large pixel field of view (FOV) may
2.0-10-3 s and 5.0-10-4s, or between 5.0-10-4s and 10-4s, or be between 0.1 and 0.2 degrees, or between 0.2 and 0.5
it might be less than 10-4s. degrees, or larger than 0.5 degrees.
US 8,809,787 B2
15 16
The device's relatively large number of pixels may be (j) eliminating from the selection those of the pixels, whose
between 10,000 and 100,000 pixels, or between 100,000 and signals are not accompanied by an acoustic signal with a
1,000,000 pixels, or more than 1,000,000, pixels. signature Substantially similar to the signature of a shot.
The device's relatively large field of view may be between The processing unit for use in detection of a muzzle flash
30 and 50 degrees, or between 50 and 90 degrees, or larger event may include at least one of:
than 90 degrees. (a) an analog processing unit, adapted to apply analog
The device's spectral range may include a part or the whole processing to time dependence of the pixel signals, so as to
of the trough of low atmospheric light transmission situated generate Substantially an estimate of likelihood that a pixel
around 1.15um. As well, it may include a part or the whole of detected muzzle flash;
the trough of low atmospheric light transmission situated 10 (b) a parallel processing unit, adapted to apply parallel
around 1.39 um, and/or the trough of low atmospheric light processing to time dependence of the pixel signals, so as to
generate Substantially an estimate of likelihood that a pixel
transmission situated around 1.9 Lim, and/or the trough of low detected muzzle flash;
atmospheric light transmission situated around 2.7 um. (c) a digital processing unit, adapted to apply digital pro
The device's PDA may beat least partially sensitive within 15 cessing to time dependence of the pixel signals, so as to
the NIR spectrum. the PDA may be sensitive only within the generate Substantially an estimate of likelihood that a pixel
NIR spectrum. The PDA may be at least partially sensitive detected muzzle flash;
within the SWIR spectrum. The PDA may be sensitive only (d) at least two processing stages, a later of the stages
within the SWIR spectrum. applying a processing for the selection of the candidate pixels
The device's PDA may be a CMOS PDA. The PDA may be to a smaller number of the pixels than an earlier of the stages;
an intracavity PDA. (e) at least two processing stages, a later of the stages
The PDA integration time may be between 10-2 s and applying, to the pixels, a processing for the selection of the
5.0-10-3 s, or between 5.0-10-3 s and 2.0-10-3 s, or between candidate pixels, using more processing time per pixel than
2.0-10-3 s and 5.0-10-4s, or between 5.0-10-4s and 10-4s, or an earlier of the stages.
it might be less than 10-4s. 25 In the processing unit, the eliminating of background por
The device's PDA dead time may be lower thana millisec tions may include determining scene motion vectors.
ond. In the processing unit, the eliminating relatively slowly
The processing unit may be in-pixel. changing portions from the pixel signals may include at least
The device may include a compressing unit accommodated one differencing of the pixel signals. The differencing may be
to compress a replica of the pixel signals. 30 second order differencing of the pixel signals.
The device may be configured to have a relatively low In the processing unit, selecting candidate pixels by com
weight. paring an estimate of likelihood that a pixel detected muzzle
According to another broad aspect of the invention, there is flash with a threshold may include selecting a pulse in the
form of a peak. The peak may be of a predetermined duration.
provided a processing unit for use in detection of a muzzle 35 The analog processing unit may include a circuit config
flash event. The processing unit may be adapted to process ured for performing at least one differencing of the pixel
pixel signals originating from a PDA, and adapted for use in signals. The circuit may be configured for performing the
selection of candidate pixels, whose signals are substantially second order differencing of the pixel signals.
similar to the temporal-spatial signature of muzzle flash. The The analog processing unit may include a matched filter,
processing unit may be also adapted to perform at least one of 40 accommodated on a pass of the pixels signals, the matched
the following: filter being adapted to match the temporal signature of muzzle
(a) eliminating spatial background portion from the pixel flash.
signals: The analog processing unit, adapted to apply analog pro
(b) eliminating relatively slowly changing portions from cessing to time dependence of the pixel signals so as to
the pixel signals; 45 generate Substantially an estimate of likelihood that a pixel
(c) selecting pixel signals’ portions Substantially similar to detected muzzle flash may be adapted to generate the Sub
the temporal signature of muzzle flash; stantial estimate as a Voltage potential at an output of the
(d) selecting the candidate pixels by comparing a Substan analog processing unit.
tial estimate of likelihood that a pixel detected muzzle flash The processing unit may use common thresholds within at
with a threshold common for a plurality of operating pixels; 50 least two groups of adjacent pixels, the thresholds differing
(e) selecting the candidate pixels by comparing an estimate between the groups.
of likelihood that a pixel detected muzzle flash with a thresh According to a broad aspect of the invention, there is pro
old obtained using replicas of the pixel signals; vided a device for use in detection of a muzzle flash event, the
(f) selecting Substantially the same number of the candi device including a PDA, relatively highly sensitive in at least
date pixels from Subsequent frames; 55 a portion of the NIR and SWIR spectrum, and a filter of
(g) selecting, as the candidate pixels, those of the pixels, electromagnetic radiation, selectively passing Substantially a
whose signals together with their vicinity pixels signals spectral range corresponding to relatively low light transmis
present Substantially the spatial signature of muzzle flash; sion in atmosphere.
(h) eliminating from the selection those of the pixels, According to a broad aspect of the invention, there is pro
whose signals together with their vicinity pixels signals 60 vided a device for use in detection of a muzzle flash event, the
present Substantially the temporal-spatial signature of a mov device including a PDA, having a cutoff wavelength lower
ing light source: than 3 microns, and a readout circuit for the PDA, the circuit
(i) confirming, as the candidate pixels, those of the pixels, being configured and operable to sample each operating pixel
whose signals together with their vicinity pixels signals, of the PDA more than 500 times per second.
obtained after their initial selection of the candidate pixels, 65 According to a broad aspect of the invention, there is pro
increase an estimate of likelihood that a pixel detected muzzle vided a device for use in detection of a muzzle flash event, the
flash. device including a PDA, having a cutoff wavelength shorter
US 8,809,787 B2
17 18
than 3 microns, and a processing unit adapted to detect Photo Detector Array (PDA) sensitive in at least a portion of
muzzleflash events in an output of said imaging arrangement. the NIR and SWIR spectrum and multiplexing pixel signals
According to a broad aspect of the invention, there is pro of the PDA into at least two replicas.
vided a device for use in detection of a muzzle flash event, the According to a broad aspect of the invention there is pro
device including a PDA, sensitive in at least a portion of the 5 vided a method for use in detection of a muzzle flash event,
NIR and SWIR spectrum; a filter of electromagnetic radiation the method including sensing, by a Photo Detector Array
selectively passing Substantially a spectral range correspond (PDA), electromagnetic radiation, passed through a filter of
ing to relatively low light transmission in atmosphere; the electromagnetic radiation selectively passing a spectral range
PDA being configured to operate with a relatively small inte of low atmospheric transmission, and multiplexing pixel sig
gration time for sensing electromagnetic radiation. The 10
nals of the PDA into at least two replicas. The method may
device may include a processing unit adapted to process pixel include recording a replica of the pixel signals into a memory.
signals originating from the PDA, the unit being adapted for According to a broad aspect of the invention there is pro
use in selection of candidate pixels, whose signals are Sub vided a method for use in detection of a muzzle flash event,
stantially similar to the temporal-spatial signature of muzzle
flash. 15 the method including sensing, by a Photo Detector Array
According to a broad aspect of the invention there is pro (PDA) sensitive in at least a portion of the NIR and SWIR
vided a method for use in detection of a muzzle flash event, spectrum, electromagnetic radiation, passed through a filter
the method including sensing, by a Photo Detector Array of electromagnetic radiation selectively passing in the portion
(PDA) which is sensitive in at least a portion of the NIR and a spectral range of low atmospheric transmission, the PDA
SWIR spectrum, electromagnetic radiation, passed through a having a fill factor larger than 60%. The fill factor may be
filter of electromagnetic radiation selectively passing in the between 60% and 75%. The fill factor may be between 75%
portion a spectral range of low atmospheric transmission, the and 90%. The fill factor may be higher than 90%.
sensing having an integration time shorter than a duration of According to a broad aspect of the invention there is pro
the muzzle flash event. vided a method for use in detection of a muzzle flash event,
According to a broad aspect of the invention there is pro 25 the method including sensing electromagnetic radiation by a
vided a method for use in detection of a muzzle flash event, Photo Detector Array (PDA) sensitive in at least a portion of
the method including sensing, by a Photo Detector Array the NIR and SWIR spectrum, and detecting the muzzle flash
(PDA), sensitive in at least a portion of the NIR and SWIR in output of the PDA by applying a signal processing to pixel
spectrum, electromagnetic radiation, passed through a filter signals of the PDA.
of electromagnetic radiation selectively passing in the portion 30 According to a broad aspect of the invention there is pro
a spectral range of low atmospheric transmission, the sensing vided a method for use in detection of a muzzle flash event,
having an integration time shorter than 10-2 s. the method including sensing by a Photo Detector Array
According to a broad aspect of the invention there is pro (PDA) electromagnetic radiation passed through a filter of
vided a method for use in detection of a muzzle flash event, electromagnetic radiation selectively passing a spectral range
the method including sensing by a Photo Detector Array 35 of low atmospheric transmission, and detecting the muzzle
(PDA) electromagnetic radiation, passed through a filter of flash in output of the PDA by applying a signal processing to
electromagnetic radiation selectively passing a spectral range pixel signals of the PDA.
of low atmospheric transmission and focused on the PDA, a According to a broad aspect of the invention there is pro
per pixel field of view (FOV) of the PDA being larger at the vided a method for use in detection of a muzzle flash event,
focusing distance than the muzzle flash. 40 the method including sensing electromagnetic radiation by a
The focusing distance may be larger than 50 m. Photo Detector Array (PDA) sensitive in at least a portion of
According to a broad aspect of the invention there is pro the NIR and SWIR spectrum, and selecting candidate pixels
vided a method for use in detection of a muzzle flash event, by applying a stage of processing to pixel signals of the PDA,
the method including sensing by a Photo Detector Array the stage being configured to provide a stage data rate
electromagnetic radiation, passed through a filter of electro 45 decrease factor Smaller than a data rate decrease factor
magnetic radiation selectively passing a spectral range of low between an initial data rate of the PDA and an alarm rate
atmospheric transmission and focused on the PDA, a per pixel benchmark of 100 alarm/s. The benchmark is an estimate of a
field of view (FOV) of the PDA being larger than 0.1 degrees. maximum shooting rate that may be caused by a single
The Photo Detector Array may be sensitive in at least a shooter. This estimate is not limiting for embodiments which
portion of the NIR and SWIR spectrum. 50 are not specifically adjusted in view of this benchmark.
According to a broad aspect of the invention there is pro According to a broad aspect of the invention there is pro
vided a method for use in detection of a muzzle flash event, vided a method for use in detection of a muzzle flash event,
the method including sensing electromagnetic radiation by a the method including sensing by a Photo Detector Array
Photo Detector Array (PDA) sensitive in at least a portion of (PDA) electromagnetic radiation passed through a filter of
the NIR and SWIR spectrum, the sensing recurring with a 55 electromagnetic radiation selectively passing a spectral range
dead time shorter than a duration of the muzzle flash. of low atmospheric transmission, and selecting candidate
The dead time may be shorter than the one tenth of the pixels by applying a stage of processing to pixel signals of the
duration. PDA, the stage being configured to provide a stage data rate
According to a broad aspect of the invention there is pro decrease factor Smaller than a data rate decrease factor
vided a method for use in detection of a muzzle flash event, 60 between an initial data rate of the PDA and an alarm rate
the method including sensing electromagnetic radiation by a benchmark of 100 alarms/s.
Photo Detector Array (PDA) sensitive in at least a portion of The stage data rate decrease factor may be smaller than 20
the NIR and SWIR spectrum, the sensing recurring with a multiplied by a square root of the data rate decrease factor
dead time shorter than one millisecond. between the initial data rate of PDA and the alarm rate bench
According to a broad aspect of the invention there is pro 65 mark.
vided a method for use in detection of a muzzle flash event, The processing may include comparing pixel signals of the
the method including sensing electromagnetic radiation by a PDA with a temporal-spatial signature of the muzzle flash.
US 8,809,787 B2
19 20
The processing may include eliminating spatial back later stage of the processing. The later stage of processing
ground portion from the pixel signals. The processing may may utilize the candidate pixel’s signal and candidate pixels
include eliminating Substantially slowly changing portions vicinity pixel’s signals.
from the pixel signals. The eliminating may include at least The processing may include Suppressing selection into
one differencing of the pixel signals. The eliminating may candidate pixels of a pixel whose signal is not accompanied
include second order differencing of the pixel signals. by a detection of an acoustic signal with an acoustic signature
The processing may include selecting pixel signals’ por Substantially similar to the acoustic signature of the muzzle
tions substantially similar to a temporal signature of the flash-causing event.
muzzle flash. The selecting may include selecting a pulse The method may include cooling of the PDA. The sensing
having a form of a peak. The peak may be of a predetermined 10 may be performed at a temperature between -30° C. and a
duration. PDA ambient temperature. The sensing may be performed at
The processing may include an analog processing of time a temperature higher than a PDA ambient temperature.
dependence of the pixel signals. The analog processing may The sensing may be performed without cooling the PDA.
include at least one differencing of the pixel signals. The The method may include utilizing for the detection a power
analog processing may include second order differencing of 15 Supply allowing more than 1 hour of the detection.
the pixel signals. The analog processing may include passing According to a broad aspect of the invention there is pro
the pixels signals through a matched filter, adapted to match vided a method for use in detection of a muzzle flash event,
the temporal signature of muzzle flash. The analog processing the method including sensing, by a Photo Detector Array
may provide a Voltage potential at an output, the Voltage (PDA), sensitive in at least a portion of the NIR and SWIR
potential being indicative of an estimate of likelihood of a spectrum, electromagnetic radiation, passed through a filter
detection of the muzzle flash. of electromagnetic radiation selectively passing in the portion
The processing may include a parallel processing of time a spectral range of low atmospheric transmission, the method
dependence of the pixel signals. including utilizing for the detection a power Supply allowing
The analog processing may be adapted to generate Substan from 1 to 24 hours of the detection.
tially estimates of likelihood that a respective to the estimate 25 The power supply may allow more than 6 hours of the
pixel has detected the muzzle flash. detection.
The parallel processing may be adapted to generate Sub The power supply may allow less than 12 hours of the
stantially estimates of likelihood that a respective to the esti detection.
mate pixel has detected the muzzle flash. The method may include utilizing for the detection a power
The processing may include a digital processing of time 30 supply of less than 10 Watts output power.
dependence of the pixel signals. The method may include utilizing for the detection only
The digital processing may be adapted to generate substan equipment portable by a human during the detection. A
tially estimates of likelihood that a respective to the estimate weight of the equipment may be less than 3 kg.
pixel has detected the muzzle flash. The power supply may allow more than 6 hours of the
The processing may include selecting candidate pixels by 35 detection.
comparing, for a plurality of pixels, Substantial estimates of The utilized sensing at least for a part may be performed
likelihood that a respective to the estimate pixel has detected within the NIR spectrum. The utilized sensing at least for a
the muzzle flash with a likelihood threshold being common part may be performed within the SWIR spectrum. The PDA
for a plurality of pixels. The threshold may be the same within may have a sensitivity maximum at a wavelength longer than
each of two or more groups of adjacent pixels and different 40 3 microns and a sensitivity cut-off at a wavelength shorter
between the groups. than 5 microns. The PDA may have a sensitivity maximum at
The processing may include selecting candidate pixels by a wavelength shorter than 3 microns. The PDA may have a
comparing, for at least one pixel, an estimate of likelihood sensitivity cut-offat a wavelength shorter than 5 microns. The
that the pixel has detected the muzzle flash with a likelihood PDA may have a sensitivity cut-off at a wavelength between
threshold, obtained using a replica of the pixel signals. 45 1.4 um and 1.65um. The PDA may have a sensitivity cut-off
The method may include selecting Substantially the same at a wavelength 1.65 um and 1.8 um. The PDA may have a
number of candidate pixels from subsequent PDA frames. sensitivity cut-off at a wavelength between 1.8 um and 2.5
The method may include utilizing at least two processing um. The PDA may have a sensitivity maximum at a wave
stages, a later of the stages applying a processing selecting length longer than 0.75 microns.
candidate pixels to a smaller number of the PDA pixels than 50 The PDA may have a region of a predominant sensitivity
an earlier of the stages. fully within the NIR/SWIR range, the region being a region
The method may include using at least two processing where the sensitivity is higher than 20% of a maximum
stages, a later of the stages selecting candidate pixels by PDA's sensitivity. The predominant sensitivity may be
utilizing more processing time per its candidate pixel than an defined as higher than 35% of a maximum PDA’s sensitivity.
earlier of the stages. 55 The predominant sensitivity may be defined as higher than
The method may include selecting into candidate pixels of 50% of the maximum PDA's sensitivity. The predominant
pixels whose signals together with their vicinity pixels sig sensitivity may be defined as higher than 70% of a maximum
nals present Substantially a spatial signature of the muzzle PDA's sensitivity.
flash. The sensing may be substantially within a range of low
The processing may include Suppressing selection into 60 atmospheric light transmission at least partially including the
candidate pixels of a pixel whose signal together with its trough situated around 1.15um (micron). The sensing may be
vicinity pixels signals present Substantially a temporal-spa Substantially within a range of low atmospheric light trans
tial signature of a Substantially moving light source. mission at least partially including a trough extending from
The processing may include at least two stages, the pro 1.34 um to 1.50 um. The sensing may be substantially within
cessing at a later stage including confirming selection into 65 a range of low atmospheric light transmission at least partially
candidate pixels of a candidate pixel whose likelihood esti including a trough extending from 1.80 um to 2.00 um. The
mate of having detected the muzzle flash has increased at the sensing may be substantially within a range of low atmo
US 8,809,787 B2
21 22
spheric light transmission at least partially including a trough According to a broad aspect of the invention there is pro
extending from 2.50 um to 2.90 um. vided a device for use in detection of a muzzle flash event, the
The method may include compressing a replica of pixel device including a Photo Detector Array (PDA) sensitive in at
signals of the PDA before the recording. least a portion of the NIR and SWIR spectrum, a filter of
According to a broad aspect of the invention there is pro electromagnetic radiation selectively passing in the portion a
vided a device for use in detection of a muzzle flash event, the spectral range of low atmospheric transmission, the PDA
device including a Photo Detector Array (PDA), sensitive in having a fill factor larger than 60%.
at least a portion of the NIR and SWIR spectrum, and a filter According to a broad aspect of the invention there is pro
of electromagnetic radiation selectively passing in the portion 10
vided a device for use in detection of a muzzle flash event, the
a spectral range of low atmospheric transmission, the PDA device including a Photo Detector Array (PDA) sensitive in at
having an integration time shorter than a duration of the least a portion of the NIR and SWIR spectrum, and a process
muzzle flash event. ing unit adapted to detect the muzzle flash in output of the
According to a broad aspect of the invention there is pro PDA by applying a signal processing to pixel signals of the
vided a device for use in detection of a muzzle flash event, the 15
PDA.
device including a Photo Detector Array (PDA), sensitive in According to a broad aspect of the invention there is pro
at least a portion of the NIR and SWIR spectrum, and a filter vided a device for use in detection of a muzzle flash event, the
of electromagnetic radiation selectively passing in the portion device including a Photo Detector Array (PDA), a filter of
a spectral range of low atmospheric transmission, the sensing electromagnetic radiation selectively passing a spectral range
having an integration time shorter than 10-2 s. of low atmospheric transmission, and a processing unit
According to a broad aspect of the invention there is pro adapted to detect the muzzle flash in output of the PDA.
vided a device for use in detection of a muzzle flash event, the According to a broad aspect of the invention there is pro
device including a Photo Detector Array (PDA), a filter of vided a device for use in detection of a muzzle flash event, the
electromagnetic radiation selectively passing a spectral range device including a Photo Detector Array (PDA), sensitive in
of low atmospheric transmission, and optics adapted to focus 25 at least a portion of the NIR and SWIR spectrum, and a
at least the spectral range on the PDA, a per pixel field of view processing unit including at least one stage of processing
(FOV) of the PDA being larger at the focusing distance than adapted to select candidate pixels by applying the stage to
the muzzle flash. pixel signals of the PDA, the stage being configured to pro
According to a broad aspect of the invention there is pro vide a stage data rate decrease factor Smaller than a data rate
vided a device for use in detection of a muzzle flash event, the 30 decrease factor between an initial data rate of the PDA and an
device including a Photo Detector Array, a filter of electro alarm rate benchmark of 100 alarms/s.
magnetic radiation selectively passing a spectral range of low According to a broad aspect of the invention there is pro
atmospheric transmission, and optics adapted to focus at least vided a device for use in detection of a muzzle flash event, the
the spectral range on the PDA, a per pixel field of view (FOV) device including a Photo Detector Array (PDA), a filter of
of the PDA being larger than 0.1 degrees. 35 electromagnetic radiation selectively passing a spectral range
A device for use in detection of a muzzle flash event, the of low atmospheric transmission, and a processing unit
device including a Photo Detector Array (PDA) sensitive in at including at least one stage of processing adapted to select
least a portion of the NIR and SWIR spectrum, the device candidate pixels by applying the stage to pixel signals of the
adapted to operate with a dead time shorter than a duration of PDA, the stage being configured to provide a stage data rate
the muzzle flash. 40 decrease factor Smaller than a data rate decrease factor
According to a broad aspect of the invention there is pro between an initial data rate of the PDA and an alarm rate
vided a device for use in detection of a muzzle flash event, the benchmark of 100 alarms/s.
device including a Photo Detector Array (PDA) sensitive in at The stage data rate decrease factor may be smaller than 20
least a portion of the NIR and SWIR spectrum, the device multiplied by a square root of the data rate decrease factor
adapted to operate with a dead time shorter than one milli 45 between the initial data rate of the PDA and the alarm rate
second. benchmark.
The device may include a shutter, which may be an elec The processing unit may be adapted to compare pixel Sig
tronic shutter. nals of the PDA with a temporal-spatial signature of the
The device may include a shutter controller for operating a muzzle flash.
shutter with the selected dead time. 50 The processing unit may be adapted to eliminate a spatial
The shutter controller may utilize a rolling shutter scheme background portion from the pixel signals.
for operating the shutter. The processing unit may be adapted to generate scene
According to a broad aspect of the invention there is pro motion vectors.
vided a device for use in detection of a muzzle flash event, the The processing unit may be adapted to eliminate Substan
device including a Photo Detector Array (PDA) sensitive in at 55 tially slowly changing portions from the pixel signals. The
least a portion of the NIR and SWIR spectrum and a multi processing unit may be configured to perform the elimination
plexer, adapted to divide pixel signals originating from the using at least one differencing of the pixel signals. The pro
PDA into at least two replicas. cessing unit may be configured to perform the elimination
According to a broad aspect of the invention there is pro using a second order differencing of the pixel signals.
vided a device for use in detection of a muzzle flash event, the 60 The processing unit may be adapted to select pixel signals
device including a Photo Detector Array (PDA), a filter of portions Substantially similar to a temporal signature of the
electromagnetic radiation selectively passing a spectral range muzzle flash.
of low atmospheric transmission, and a multiplexer, adapted The processing unit may have a part that is in-pixel.
to divide pixel signals originating from the PDA into at least The processing unit may include an analog processing unit
two replicas. 65 adapted to process time dependence of the pixel signals. The
The device may include a memory for storing a replica of analog processing unit may include a circuit adapted to per
the pixel signals. form at least one differencing of the pixel signals. The analog
US 8,809,787 B2
23 24
processing unit may include a circuit adapted to perform The device may be configured without a cooler for the
second order differencing of the pixel signals. PDA.
The analog processing unit may include a matched filter, The device may include a power Supply allowing more than
adapted to match the temporal signature of muzzle flash. 1 hour of the detection.
The analog processing unit may be adapted to provide a 5 According to a broad aspect of the invention there is pro
Voltage potential at outputs, the Voltage potential being vided a device for use in detection of a muzzle flash event, the
indicative of an estimate of likelihood of a detection of the device including a Photo Detector Array (PDA), sensitive in
muzzle flash.
The processing unit may include a parallel processing unit at of
least a portion of the NIR and SWIR spectrum, and a filter
electromagnetic radiation selectively passing in the portion
for processing time dependence of the pixel signals. 10 a spectral range of low atmospheric transmission, the device
The analog processing unit may be adapted to generate
substantially estimates of likelihood that a respective to the including
detection.
a power Supply allowing from 1 to 24 hours of the
estimate pixel has detected the muzzle flash. The power supply may allow more than 6 hours of the
The parallel processing unit may be adapted to generate
substantially estimates of likelihood that a respective to the 15 detection.
estimate pixel has detected the muzzle flash. The power supply may allow less than 12 hours of the
The processing unit may include a digital processing unit detection.
processing time dependence of the pixel signals. The device may include a power supply of less than 10
The digital processing unit may be adapted to generate Watts output power.
substantially estimates of likelihood that a respective to the 20 The device may be portable by a human during the detec
estimate pixel has detected the muzzle flash. tion.
The processing unit may be adapted to select candidate The PDA may be a CMOS PDA. The PDA may be an
pixels by comparing, for a plurality of pixels, Substantial intracavity PDA.
estimates of likelihood that a respective to the estimate pixel The device may include a compressing unit adapted to
has detected the muzzle flash with a likelihood threshold 25 compress a replica of pixel signals of the PDA before the
being common for a plurality of pixels. recording them into a memory.
The threshold may be the same within each of two or more According to a broad aspect of the invention there is pro
groups of adjacent pixels and different between these groups. vided a device for use in muzzle flash detection, the device
The processing unit may be adapted to select candidate including a Photo Detector Array (PDA), having a cutoff
pixels by comparing, for at least one pixel, an estimate of 30 wavelength shorter than 3 microns, and a processing unit
likelihood that the pixel has detected the muzzle flash with a adapted to detect muzzle flash events in an output of the PDA.
likelihood threshold, obtained using a replica of the pixel The device may include a filter of electromagnetic radia
signals. tion selectively passing in the PDA's sensitivity band a spec
The processing unit may be adapted to select Substantially tral range of low atmospheric transmission.
the same number of candidate pixels from subsequent PDA 35 The filter may pass less than 50% of energy of wavelength
frames. being outside the spectral range of low atmospheric transmis
The device may include at least two processing stages, a sion and sensed by the PDA.
later of the stages configured to apply a processing selecting The filter may pass less than 25% of energy of wavelengths
candidate pixels to a smaller number of the PDA pixels than being outside the spectral range of low atmospheric transmis
an earlier of the stages. 40 sion and sensed by the PDA.
The device may include at least two processing stages, a The filter may pass less than 10% of energy of wavelengths
later of the stages configured to select candidate pixels by being outside the spectral range of low atmospheric transmis
utilizing more processing time per its candidate pixel than an sion and sensed by the PDA.
earlier of the stages. The filter may pass less than 2% of energy of wavelengths
The processing unit may be adapted to selecting into can- 45 being outside the spectral range of low atmospheric transmis
didate pixels those pixels whose signals together with their sion and sensed by the PDA.
vicinity pixels signals present Substantially a spatial signa The device may have at at least one wavelength of the
ture of the muzzle flash. spectral range a sensitivity being between 50% and 75% of
The processing unit may be adapted to Suppress selection the sensitivity of the PDA.
into candidate pixels of a pixel whose signal together with its 50 The device may have at at least one wavelength of the
vicinity pixels signals present Substantially a temporal-spa spectral range a sensitivity larger than 75% of the sensitivity
tial signature of an unsuitably moving light source. of the PDA.
The processing unit may include at least two processing According to a broad aspect of the invention there is pro
stages, a later stage of the stages adapted to confirm selection vided a processing unit for use in detection of a muzzle flash
into candidate pixels of a candidate pixel whose likelihood 55 event, the processing unit being adapted to process pixel
estimate of having detected the muzzle flash has increased at signals originating from a PDA and to generate Substantially
the later stage of the processing. likelihoods of muzzle flash detection for pixels of the PDA,
The later stage of processing may utilize the candidate the processing unit including a multiplexer dividing the pixel
pixel’s signal and candidate pixel’s vicinity pixels signals. 222 signals between at least two branches.
The processing unit may be adapted to Suppress selection 60 According to a broad aspect of the invention, there is pro
into candidate pixels of a pixel whose signal is not accompa vided a device for use in detection of a muzzle flash event, the
nied by a detection of an acoustic signal with an acoustic device being Substantially as described in the patent applica
signature Substantially similar to the acoustic signature of the tion with reference to the specification.
muzzle flash-causing event. According to a broad aspect of the invention, there is pro
The device may include a cooler for the PDA. The cooler 65 vided a processing unit for use in detection of a muzzle flash
may be of a kind enabling sensing at a temperature lower than event, the processing unit being Substantially as described in
a PDA ambient temperature. the patent application with reference to the specification.
US 8,809,787 B2
25 26
According to a broad aspect of the invention, there is pro particular, a detection of those signals portions, which are due
vided a method being substantially as described in the patent to the muzzle flash, may be carried out, with a certain detec
application with reference to the specification. tion probability and false alarm rate. In the present example,
the detection includes a parallel processing, filtering signal
BRIEF DESCRIPTION OF THE DRAWINGS portions varying as flash-type intensity (in the selected wave
length range) with time. This parallel processing may consti
To further clarify the above and other advantages and fea tute a first layer of processing in the inventors’ technique.
tures of the present invention, and to further show how it may The inventors' technique may have a stage at which it
be carried out in practice, an additional, at times more par determines an estimate of likelihood that an arbitrary pixel
ticular, description of the invention and invention features 10 (e.g. each of the pixels) has detected a muzzle flash event.
will be rendered in the below detailed description, at times This estimate actually can be the output of the parallel pro
with reference to the appended drawings. It is appreciated that cessing.
these drawings, when depict only particular embodiments of The inventors' technique may determine a 'suspicious’
the invention, are not to be considered limiting of its scope. pixel (or pixels) for which the estimated likelihood of detec
Hence, the invention will continued to be described and 15 tion of a muzzle flash is higher than for others (this can be
explained with additional specificity and detail through the done by a pixel selection unit, such as CFAR). The method
use of the accompanying drawings in which: may also have a step at which it determines a pixel (or pixels)
FIG. 1A is an example flow chart of the detection method, which likelihood is higher than a certain threshold. An alarm
according to the to the inventors' technique; signal intended to informan operator (or operating utility) or
FIG. 1B is an example of a possible organization of a a user about the possible muzzle flash event may then be
detection system according to the inventors' technique; produced. The alarm signal may be a simple signal, e.g. a
FIGS. 2A and 2B schematically illustrate two examples of Sound tone or a light flash, or it may be a composite signal
a detection system, according to the inventors’ technique; carrying Such data as for example pixel position, PDA orien
FIG. 2C shows the sensitivity curves of various exemplary tation, determined likelihood, direction to the flash. The data
materials used in photodetectors; 25 can be presented in a form perceived by another device or by
FIGS. 3A-3D exemplify some possible configurations of humans. The alarm signal may be recorded on a memory
the processing system, according to the inventors’ technique, carrier.
capable of use in the detection system, according to the inven The inventors’ technique may include processing aimed at
tors' technique; reestimation of the likelihoods for a portion of pixels. The
FIGS. 4A and 4B exemplify the operation of the analog 30 likelihoods may be reestimated by a second layer processing
processing unit, according to the inventors' technique; using signals obtained from pixels being close to the Suspi
FIGS.5A-5C illustrate the principles underlying the use of cious pixels and/or signals obtained at times being relatively
an optical filter in the technique of the inventors; and close to the moment for which the reestimation is done.
FIG. 6 schematically illustrates an example of a weapons In particular, the second layer of processing may perform
firing detection system organized according to the inventors 35 the following. For each of the suspicious pixels the second
technique. layer of processing may reestimate the likelihood that this
pixel has detected a muzzle flash. To this end the second layer
DETAILED DESCRIPTION OF THE PREFERRED of processing may, for example, check whether the Suspicious
EMBODIMENTS pixel belongs to a two-dimensional (2D) projection of trajec
40 tory of an object crossing the field of view and generating a
The present invention provides a novel technique for use in strong light signal. In such case the likelihood corresponding
detection of short events, for example of a gunshot event to this pixel may be decreased, because the signal could be
including a muzzle flash event. produced by a strong Sunlight reflector or artificial light
FIG. 1A shows a flow chart of the main steps in an example Source, e.g. by a bird crossing the field of view or a car light
of the detection method of the invention. As shown, near 45 or a Sun glint. A tracking test aimed at discriminating pixels
infrared (NIR) and/or short wave infrared (SWIR) electro grouping in trajectories may be based for example on the
magnetic radiation, possibly including radiation emitted by Hough Transform. For another example of the operation of
an occurred muzzle flash, is collected from a field of view the second level processing, the likelihood corresponding to a
(FOV). The radiation is spectrally filtered for detection of Suspicious pixel may be increased if this pixel has neighbor
Substantially a spectral range corresponding to relatively low 50 ing pixels which signals or likelihoods are indicative of an
light transmission in atmosphere. As it has been found by the existence of a pixel group having a muzzle flash characteristic
inventors, the muzzle flash radiation components being in shape. For example, this characteristic shape can be similar to
Such spectral range can provide a relatively high useful signal the infinity sign. For yet another example of the operations
and signal-to-noise ratio, though they carry only a portion of performable by the second level processing, the likelihood
overall muzzle flash intensity and are relatively short-exist 55 corresponding to a Suspicious pixel may be adjusted accord
ing. Also, it should be understood that only a portion of the ing to the spatial and temporal characteristic of the pixels
collected light can be due to the muzzle flash event, i.e. can be signals. Such adjustment may account for the possibility of
useful. The radiation of the chosen spectral range is detected fire bursts. If the second layer of processing receives two or
with an appropriate photodetector array (PDA). The PDA more Suspicious flashes separated by a certain characteristic
integration time may be between 10-2 s and 5.0-10-3 s, or 60 time period and possibly by few pixels, it can increase the
between 5.0-10-3 s and 2.0-10-3 s, or between 2.0-10-3 s and likelihoods of these events being muzzled flashes, because
5.0-10-4s, or between 5.0-10-4s and 10-4s, or it might be less they could be produced by e.g. a machine gun. On the con
than 10-4s. As a result, outputs of the PDA provide electrical trary, detection of aperiodic suspicious events separated by
signals, some of which are indicative of the muzzle flash, if it small time periods may be indicative of the absence of muzzle
has occurred. 65 flashes, because there would be no enough time for weapon
The example of the detection method optionally includes recharge. The above-described processing is facilitated if sig
also other steps some of which are shown in FIG. 1A. In nals produced by the PDA and/or first layer of processing are
US 8,809,787 B2
27 28
stored in a memory for a time period in which they may be level-2 may utilize for connection a pixel selecting utility 40,
useful for the second and other layers of processing. The data such as the CFAR utility, so that a number of suspicious
from the PDA and/or first layer when they are prepared for events does not overflow layer-2 processing capacity. For
further use can be sampled or transformed, e.g. partially each of the Suspicious pixels the layer-2 processing may need
averaged, into less broadband data. The data also can be to obtain the values of its neighboring pixels, several frames
compressed (without loss of information) and stored in the before and after the suspicious event occurred. To this end
memory in the compressed form. outputs of the PDA or any level (level-0, level-1 and/or level
The first layer of processing thus may work with the second 2) processing unit may be stored in memory utility 46.
layer of processing in a Constant False Alarm Rate regime The pixel-selecting utility 40 can prepare a list of candidate
(CFAR regime): a constant stream of Suspicious (candidate) 10 pixels for the layer-2. To this end, this utility can compare the
events from the first layer may be selected and processed by likelihoods calculated by the layer-1 or it can generate the
the second layer. The second layer is designed to handle a likelihoods based on the output of the layer-1. It also may be
predetermined constant flow of Suspicious events (e.g. 10 or configured to connect with level-0 and/or memory (these
20 perframe). Due to the significant data reduction performed connections are not shown in FIG. 1B), because it may use
by the first layer (e.g. from ~100,000 pixels per frame to ~10 15 historical or spatial data to calculate temporal and spatial
Suspicious pixels per frame), the second layer has signifi statistics for pixels, for example statistics on luminance level
cantly more time per pixel to analyze the Suspicious events. A in the vicinity of the pixel of the candidate event and/or before
Suspicious eventis confirmed and an actual alarm is produced and/or after the occurrence of the candidate event in order to
if the reestimated likelihood of that the corresponding suspi evaluate the spatial statistical significance of the event.
cious pixel has detected a muzzle flash event is larger than a Indeed, the utility may benefit from taking into account infor
certain threshold. mation gathered after the candidate event had occurred, if it is
FIG. 1B schematically illustrates an example of a system configured to use for reestimation data obtained after a can
50 configured in accordance with the technique of the inven didate event, in addition to or instead of the historical data. To
tors. System 50 includes one or more of the following: col this end, a delay of several tens to hundreds of frames between
lecting optics 5; an optical filter 10; a photodetector pixel 25 data processed in the second stage and in the first stage may be
array 20; a shutter 21, a processing system 22. Collecting utilized.
optics 5 may be configured to define a certain chosen FOV in Level-2 processing unit 45 implements the above-men
a certain chosen wavelength (sub)region (e.g. in a broader tioned second layer of processing aimed at determining Sus
NIR/SWIR region). Optical filter 10 may be configured to picious pixel(s) and reestimating the likelihood(s) that the
pass light corresponding to bands of low atmospheric trans 30 system has detected a muzzle flash. For this purpose, the
mission, particularly in the chosen NIR/SWIR wavelength second layer of processing may use signals obtained from
region. PDA20 is sensitive in the corresponding NIR/SWIR neighboring (close) pixels and/or signals obtained at close
bands. Shutter 21 may be an electronic shutter defining the time moments. Hence, level-2 processing unit combines sig
small PDA integration time, as specified above. Additionally nals received from level-1 processing unit 35 through pixel
or alternatively, shutter 21 may be configured to define a small 35 selection unit 40 and/or uses signals stored in level-1 memory
dead time. Shutter 21 may be a rolling shutter. Subsystem 22 utility 46. It also may store data that it produces (e.g. likeli
performs the processing of electric signals (i.e. of a video hood estimates) in memory utility 46 and may access the data
stream) produced by PDA 20 operated by shutter 21. earlier recorded in this memory. The processing implemented
In the present example, subsystem 22 includes a level-0 by level-1 and level-2 processing unit may be in accordance
processing unit 25, a calibration utility 24, a level-1 process 40 with the bad pixel map.
ing unit 35 performing parallel processing, a pixel selection In particular, level-2 unit may be configured to reestimate
unit 40, a level-2 processing unit 45, a compression utility 44. the muzzle flash likelihood in view of a possibility of the
and a memory utility 46. splitting of the muzzle flash projection to several close or
Level-0 processing unit 25 may apply a non-uniformity adjacent pixels (e.g. 2 or 4 pixels), which together represent
correction (i.e. correction for the non-linearity of PDA 45 the same event. In this connection, it should be noted, that a
response) to the raw signal from the pixels. Calibration utility muzzle flash projection may not lose its specific shape. Such
24 may store a bad pixels map for photodetector 20 (the bad as heart shape, balloon shape, droplet shape round shape. At
pixels can be found during calibration of the photodetector). short distances, the details of the shape may be imaged, ana
During operation of the inventors' gunshot detection system, lyzed by the level-2 unit, and used for confirmation or rejec
calibration utility 24 can eliminate the bad pixels from the 50 tion of muzzle flash event and for localization of the respec
data which are input to level-1 processing unit 35. Level-0 tive muzzle flash.
may be configured to find motion vectors. Level-2 unit may also be configured to reestimate the
Level-1 processing unit 35 performs, as described above muzzle flash likelihood in view of a possibility of various
and, more specifically, below, while referring to the examples timings of the muzzle flash relatively to the samplings, in
of FIGS. 2A-2B and 3A-3D, the data reducing processing 55 order to take into account the variability of the sampled
aimed at detection of muzzleflash, particularly through selec muzzle flash waveform. Indeed, a muzzle flash does not have
tion of muzzle-flash time-dependency in the received signals. to occur within a single integration interval, it can begin in
The level-1 may use relatively simple algorithms (such as one frame and end in a different frame.
second order derivative) to identify signals which resemble a Turning back to the level-0 processing, it may also be
muzzle flash time signature. In the parallel implementations, 60 configured to analyze the motion in the pictures, and generate
the layer-1 analyzes the time dependent signal from each frame to frame motion vectors for blocks of m times n pixels,
pixel independently of other pixels, in order to reduce com so that clutter and background can be accurately deducted at
puting power. The level-1 processing typically allows achiev the further layer(s). The frame to frame motion vectors may
ing effective data reduction of 3-6 orders of magnitude for be generated by using a block matching algorithm proved
level-2 processing or for alarm rate (the data reduction for 65 useful for motion compensation. As in MPEG applications,
level-1 can be defined as a number of pixels divided by a the generation of motion vectors may be implemented in
number of suspicious pixels in a frame). The level-1 and software or in hardware, in FPGA or ASIC. In some of the
US 8,809,787 B2
29 30
preferred adaptations (embodiments), the generation of detector is typically smaller than a similar combination of
motion vectors is performed by a programmed FPGA. In MWIR/LWIR optics and photodetectors, because diffraction
another option the generation of motion vectors is performed effect is smaller in NIR/SWIR, a larger variety of optical
at layer 2 and only for Suspicious (i.e. candidate) events. materials and manufacturing techniques is available, and
Referring to FIGS. 2A and 2B, there are schematically NIR/SWIR detectors are faster and do not require cryogenic
illustrated two examples, respectively, of a gunshot and cooling to provide fast and sensitive detection. The technique
muzzle flash detection system of the present invention. The of the inventors typically involves photodetectors having a
detection system is configured to receive radiation from a quantum efficiency of more than 20% in NIR/SWIR range
certain field of view and detect whether the received radiation used for detection.
contains a portion emitted by a muzzle flash. 10 In connection with the above, it should be noted, that many
In the example of FIG. 2A, a detection system 100 includes features of the invention are not restricted to the NIR/SWIR
a light collecting and focusing optics 5 (e.g. a lens assembly range. In particular, the inventors have considered that detec
formed by one or more lenses) in front of a photodetector tion systems utilizing other wavelength ranges can also utilize
pixel array unit 20A, a parallel processing unit 30, and a the layered architecture, the filtering of light of the low trans
control unit 38. 15 mission in atmosphere, the Suppression of candidate events
Lens assembly 5 collects light Lin from a region of interest produced by an inappropriate group of pixels, the shutting
and may have a wide field of view of tens of degrees in each scheme providing the Small dead time, the light collecting
lateral direction. It focuses light Lin onto photodetector array scheme providing relatively large field of view per pixel.
20A. The region of interest can include a range of distances up Parallel processing unit 30, capable for example of imple
to several hundred meters, e.g. a range up to 500 m from the menting level-1 processing unit 35 of FIG. 1B, is configured
lens assembly. The light focusing is applied to a predeter and operable according to the invention for filtering the elec
mined wavelength range. It should be understood that the lens trical output of the pixel array for detecting or selecting a
assembly may be operable by the control unit or manually by muzzle flash-type signal portion (e.g. for Suppressing all sig
a user to focus light from a different field of view and/or to nal portions except for possibly present flash-type signal).
focus light of a different wavelength range. Though in the 25 The processing unit may include an array of analog proces
present example the lens assembly is a constructional part of sors. These processors may be configured for simultaneous
the detection system, generally this is not a requirement. operation. In some other embodiments processing unit 30 is
Photodetector array 20A can be of any known type sensi configured to perform analog-to-digital conversion and pro
tive to a subrange of ultraviolet/visible/infrared spectra. Pho cess the digital data. The operation in the digital configuration
todetector array 20A may be based, for example, on PbS, 30 may be partially sequential. The analog-to-digital conversion
InAs, GaAs, InGaAs, MCT, PbSe, or InSb for relatively fast may be done before the level-1 processing unit, for example
and low cost NIR, SWIR or MWIR detection. In some pre already at the PDA output.
ferred embodiments the PDA is light, fast, sensitive, inexpen Detector device 100 is associated with a control unit 38,
sive, and does not require cooling and much power for opera which may or may not be a constructional part of the system.
tion. Also, in some preferred embodiments the photodetector 35 The control unit is connected to the output of the first level, for
array has a large pixel count detection array and allows accu example parallel, processing (sub)unit (via wires or wire
rate determination of a muzzle flash event location. The sen lessly) and may perform the second layer of processing. The
sitivity of various materials used in photodetectors is shown level-2 processing unit 45 and memory utility 46 of FIG. 1B
on FIG. 2C, as a dependent of wavelength. The technique of may be implemented in control unit 38.
the inventors may utilize for example InSb (Indium Anti 40 Control unit 38 is typically a computer system including
monides) and MCT (HgCdTe) at 77K for MWIR imaging, inter alia a digital signal processor 42, a memory 46, and
and InGaAs (300K) and Ex InGaAs (253K) for SWIR imag input/output utilities, generally at 48. The control unit is
1ng. configured for receiving and further processing data from
The inventors have found that the NIR/SWIR detection of analog to digital, parallel, processing unit 30; and possibly
muzzle flash can he utilized in the in-field muzzle flash detec 45 also for controlling at least some of the elements of the detec
tion applications. This is because in addition to energy in the tor device. For example it may perform the second layer of
MWIR/LWIR, the hot gasses of the muzzle-flash emit black processing and/or control the lens assembly, filter(s), PDA
body radiation energy in the NIR/SWIR range. settings (i.e. settings of the level-0 processing e.g. those input
Nevertheless, the inventors have chosen to use photodetec into level-0 processing unit from level-0 processing utility),
tors sensitive to NIR and/or SWIR light for the purposes of 50 parallel processing unit settings (threshold for pixel selec
muzzle flash detection in some embodiments of their tech tion). It also may be configured for receiving and/or providing
nique. This is because the inventors have considered that signals to an operator. Input utility 48 of control unit 38 may
photodetectors having the desired characteristics for their be configured to pass a limited number of candidate pixels to
application are more easily available in NIR/SWIR than in the second layer of processing, for example it may be config
MWIR/LWIR and possible losses in optical signal may be 55 ured to work in the CFAR regime.
overcompensated by gains from a higher sampling rate In the example of FIG. 2B, showing another preferred
(smaller integration and dead time) and frame rate of NIR/ embodiment of the muzzle flash detection system of inven
SWIR detectors; from the use of the solar blind filters which tors, detection system 100 includes a photodetector unit 20B
reduce clutter and enhance signal to clutter ratio; from the associated with a light collecting and focusing optics 5, an
increased availability of higher resolution in the NIR/SWIR 60 optical filter 10 accommodated in the optical path of light
detectors due to the higher maturity and lower cost of PDA propagating towards the pixel array 20, and an analog (in this
materials in this range; and from the relaxed need or lack of case analog parallel) processing unit 30 at the output of the
need for cooling the PDA for achieving high performance. pixel array. Filter 10 is configured and operable to enable
(Already the relaxed need is beneficial, because cryogenic detection of relatively short-living muzzle flash components,
coolers become Smaller, lighter, less power-consuming, and 65 as will be more specifically described further below with
costly with the decrease of the PDA size and heat dissipation). reference to FIGS.5A-5C. Filter 10 may be a standalone unit
For example, a combination of NIR/SWIR optics and photo or may be integrated within an intracavity detector. Photode
US 8,809,787 B2
31 32
tector unit 20B with filter 10 is sensitive to substantially a FIGS. 3B and 3C show specific but not limiting examples
spectral range corresponding to relatively low light transmis of an analog in-pixel processing unit 30P adapted to filter in
sion in atmosphere. To this end, the sensitivity of Such the output of a photodetector pixel a muzzle flash type signal
arrangement to at least one wavelength of the utilized low portion. This portion has light intensity behaving as an asym
atmospheric transmission spectral range may be between 5
metric pulse of a muzzle flash duration i.e. it has a character
10% and 30%, or between 30% and 50%, or between 50% and istic muzzle flash time variation.
75%, or larger than 75%. An average sensitivity of such In the example of FIG. 3B, analog processing unit 30P
arrangement to wavelengths outside Such range and present in processes a signal SPD from a pixel 22 of the photodetector
daylight and to which the photodetector (i.e. the imager) is array. Unit 30P includes an integrating circuit 32, a delay
fundamentally sensitive, may be smaller than 10% (or 5%, or 10
utility 34, a subtraction circuit 36, and a switch 33. Signal
1%). SPD is integrated by a circuit 32. Switch 33 is clocked with a
Also device 100 may include a phase mask 8. The phase predetermined time T (i.e. is shifted for a short time into its
mask may perform signal processing in the optical domain,
for example, subtraction of signals of two different wave ON state with increments of time T). Time T is substantially
lengths. 15 equal to a delay time of delay utility 34 and is smaller than a
Further in FIGS. 3A-3D there are exemplified some pos pre-estimated time-width (duration) of a muzzle flash pulse at
sible configurations of the processing (sub)system carrying the photodetector output (e.g. pulse P shown in FIG. 3A).
out the layer-1 in-pixel processing and, in case of FIG. 3D, Since the integrated signal is output to one input of Subtrac
also the layer-2 processing. It should be understood that the tion circuit 36 and also, through delay utility 34, to the other
layer-1 processing unit may be formed by an array of operat input of subtraction circuit 36, the time change of the pixel
ing in parallel individual processors, each of which is associ output is determined. The result of Subtraction forms an ana
ated with the detector pixel, and that the design of the parallel log signal SA, which then may be directed to the memory
processing unit is aimed at filtering muzzle flash time domain utility and/or the pixel selection utility (and then to the level-2
features in the received signal, for example at amplifying the processing unit) and/or to the control unit.
possibly present signal portion having a time-dependence of 25 In the example of FIG. 3C, analog processing unit 30P
a type producible by a muzzle flash while removing other includes a band pass filter 31, a peak detector 35, and two
signal portions. The features to be suppressed or removed switches 33A and 33B. A signal SPD from a pixel 22 of the
from the photodetector signal are those caused by noises and photodetector array is passed through band filter 31 to peak
clutter. Hence, a pulse to be filtered by the level-1 processing detector 35. The band of filter 31 is selected so as to allow
utility is that which has a well defined peak, a muzzle flash 30 passage of frequencies corresponding to Fourier transform of
type duration (for most firearms this duration is up to about a a pre-estimated muzzle flash pulse at the photodetector output
few milliseconds) and an asymmetric form typical to the (e.g. pulse P shown in FIG. 2A). Peak detector 35 outputs a
muzzle flash. Other signal portions in the photodetector peak value of this pulse at its output node 35out. Switch 33A
detector pixel output are considered as associated with clutter is a reset switch for resetting a peak detector output 35out to
or noise. 35 Zero once during a certain time interval T1. Switch 33B is an
The processing unit (e.g. 35 in FIG. 1B or 30 in FIGS. output switch for outputting the peaks to a control unit 38
2A-2B), in particular the parallel processing unit, may be an (which is not shown in this figure). This output is done right
array of matched filters. In this connection, FIG. 2A shows a before the reset of the peak detector. A sequence of the peak
pulse P created at the photodetector output in response to a values forms an analog signal SA. So designed analog pro
muzzle flash and a time response characteristics TR of the 40 cessing unit 30P can select a signal portion corresponding to
matched filter processing unit. Pulse P presents only the por a pulse of the flash-type intensity variation with time in signal
tion of a photodetector output (SPD in FIGS. 2A-2B); how SPD. Peak detector 35 enables the detection of the muzzle
ever other portions of the photodetector output are filtered out flash peaks in case of reset time T1 being shorter or longer
by the match filter. one of the uniquenesses of pulse Plies in than the time-width of a muzzleflash pulse. However, in some
its asymmetric variation with time. Pulse P grows from Zero 45 preferred configurations, the reset time T1 is selected to be
to a maximum in a first predetermined time interval, and falls shorter than the pre-estimated time-width of muzzle flash
back to Zero in a second, typically longer, predetermined time pulse, because such a selection allow for comparing a time
interval. The second time interval may be about two times width of a received pulse with the pre-estimated time-width
longer than the first time interval. The matched filter is of the muzzle flash and also because Such a selection allows
adapted to detect the uniqueness of the muzzle flash signal. 50 for resolving signals from consecutive muzzle flashes.
Various matched filter physical implementations, either ana Moreover, when time T1 is selected so as to be several
log or digital, are known per se and therefore need not be times (e.g. 10) shorter than the time-width of a muzzle flash
described in detail. It should be noted, however, that in some pulse, the analog processing unit can detect an asymmetry of
preferred embodiments the muzzle flash features are filtered a signal portion corresponding to a pulse of the flash-type
by in-pixel analog processing. Such as parallel in-pixel analog 55 intensity variation with time.
processing, or by separate digital processing following an It should be understood that the technique of the inventors
analog-to-digital converter (ADC). The in-pixel processing is not limited to the above examples of the analog processing
can achieve very high speed, while keeping the bandwidth of unit. Other configurations of analog processing unit 30P may
the output signal low (after the use of pixel-selecting unit). include various circuits, including Such known perse circuits
This allows the next level of processing to operate on a 60 as a differencing circuit, a sample-and-hold circuit, a com
reduced input rate of Suspicious events. It also allows using parator, a low pass filter, a high pass filter, an envelope detec
the specific time-dependency or shape of signal for better tor. In some preferred embodiments, the analog sampling is
distinguishing between true events and other short but differ carried out with a sampling rate less than a tenth of the
ent events. In some embodiments, the processing speed of the duration of the selected signal.
first level of processing is selected to match the PDA sam 65 Thus, the analog processing unit may be configured to have
pling speed, which is selected to produce at least several a time response allowing identification of the above-de
samples within the time of the muzzle flash. scribed asymmetric pulse P. The analog processing may be
US 8,809,787 B2
33 34
useful for facilitating a further digital processing (sampling, utilized to remove from signal SPD the features correspond
layer-2 digital processing) if it follows. ing to very short events, i.e. events much shorter than the time
Another example of the processing system of the invention width of pulse at the photodetector output. Thus received
is shown in FIG. 3D. This in pixel signal processing sub signal SPD' has a portion P1" corresponding to the muzzle
system has a first stage, in which signals obtained from dif flash and carries reduced noise and clutter.
ferent pixels are not combined, and a second stage in which Signal P1" can be output directly from the parallel process
these signals are combined. In the first stage a sensing ele ing unit or through an absolute value peak detector (not
ment SE (i.e. a pixel) sends a detected signal to a charge shown), clocked with time that in some preferred embodi
integrating transimpedance amplifier (CTIA) circuit integrat ments is shorter than the time between positive and negative
ing the signal. Signal then propagates to a variable signal 10 extremes of signal P1". If the absolute peak value detector is
detection, which deducts a Low Pass Filter (LPF) averaged used, then the extremes of signal P1' are not lost from a
signal from the current signal. The result is then input to a fast sample of output signal used by a digital processor 42 of
shift register, where the signal is processed by a program control unit 40.
mable logic, e.g. by a programmable logic implementing the Considering the above mentioned uniqueness of the
matched filter. The output of the register determines a likeli 15 muzzle flash pulse (P in FIG. 3A), this results in a uniqueness
hood that an event has occurred within a given time frame. of pulse P1": the magnitude of the positive peak of pulse P1" is
The first parallel stage of processing finishes here. The signal approximately two times larger than the magnitude of the
then is input into a slow shift register. The values in the slow negative peak, but the duration of the positive part of pulse is
shift registers are co-processed with values in slow shift reg two times shorter than of the negative part. Thus, if pulse P1’
isters of adjacent pixels. This co-processing is performed is sampled three times by the peak detector, the following
using a mask bit latch or other logic, filtering the signals for sequence of values will be generated: 0, +X, -X/2, -X/y, 0 (this
portion having a spatial signature of the event to be detected sequence presents a not limiting example). Here, X is a mag
typically for the event taking place in the middle of the mask. nitude of the positive peak of pulse P1", it depends on the
The mask scans the entire image area in order to check for an shooter-detector distance; y is some number greater than 2.
event in each of the pixels. This architecture may be used for 25 In addition to the processing unit reducing some noise and
other algorithms. Such as video motion detection, video track clutter, the system of the invention may use other means to
ing, Automatic Target Recognition. perform muzzle flash detection with high probability and low
In another example, the system of the invention can be false alarm rate. For example, a filter, processing received
organized as follows. The system uses a typical PDA and light before its detection by the pixel array (e.g. filter 10 in
digital processing. The PDA sends signals to a processing 30 FIG. 2B), can be used to increase the SNR/signal to clutter
board, where they are sampled and transformed into digital ratio.
signal, containing the signal level detected at each of the Reference is made to FIGS.5A-5C exemplifying the prin
pixel. This digital signal may be parallel. The digital signal is ciples underlying the filter selection. FIGS.5A and 5B show
input into the layer-1 processing unit. For example, the inter a wavelength dependence of the atmospheric transmission in
face between the PDA with the processing board can be a 35 the NIR and SWIR in short ranges of 100, 200, and 300 m. It
digital communication link such as the “Camera Link' stan is seen that the transmission is very low for example in a range
dard. A possible bit-stream of the digital signal can be esti around a wavelength of 1.4 Lum. According to the invention,
mated: for a 200,000 pixel detector working at 500 frame/sec the ranges of the low atmospheric transmission may be used
rate and using 12 bit encoding of the analog signal, the bit to facilitate the muzzle flash detection. Here are some of these
stream needs to be as high as 1 Gbps. The system therefore 40 ranges: 1.34 to 1.50 microns, 1.80 to 2.00 microns, 2.50 to
uses the layered processing scheme, which can allow Sam 2.90 microns, 4.1 to 4.4 microns, 5.5 to 7.3 microns. In some
pling at the selected frame rate (e.g. 500 frames/sec). The preferred embodiments the first and/or the second of these
processing at low rates may be configured without a matched ranges are used. In some preferred embodiments, the filter
filter; it may be rather configured to find at the second layer allows passage of not more than 50% of incident energy for
short events that appear in 1-2 frames and then Vanish, where 45 wavelengths being outside the low atmospheric transmission
the Suspicious events reaching the second layer are those ranges and sensed by the pixel array. In some other preferred
which prevail over a threshold, established by the second embodiments, the filter passes not more than 25%, 10%, and
layer's pixel selection utility (e.g. CFAR utility), applied to 2% of this energy.
the digital first layer. Such digital layered processing may be FIG. 5C illustrates an effect of increased signal to clutter
performed in an ASIC or a field programmable gate array 50 ratio in the ranges of low atmosphere transmission. Graph G1
(FPGA), for example of the Xilinx Vertex family. Layer-1 can shows the Sun irradiance, i.e. the intensity of Sun radiation, on
be hard coded into the FPGA, while the layer 2 algorithm (all a clear day at the sea level, Graphs G2. G3, G4 show ratios of
or part) can be exercised in Software, either based on a pro the intensity of light produced by an exemplary muzzle flash
cessor core implemented in the FPGA, or on a separate pro respectively at distances of 100, 200 and 300 m, from the
cessor, Such as control units processor. The algorithm can be 55 detector, to the sun irradiance of graph G 1. Graphs G2. G3,
partially simultaneous and partially sequential—i.e. not all, G4 have evident peaks corresponding to the ranges of low
but several, pixels are processed simultaneously. atmospheric transmission. These peaks correspond to the
Referring to FIGS. 4A and 4B, the operation performed by increased signal to clutter ratio.
the analog or digital parallel processing unit are more specifi It should be noted, that peaks in graphs G2. G3, G4 can
cally described. FIG. 4A exemplified a photodetector signal 60 decrease with an increase in the atmosphere humidity level. In
SPD to be processed by the analog processing unit. Signal this case the spectral range may be widened to include wave
SPD is composed of clutter and various noise portions, and lengths of Smaller atmospheric absorption.
has a muzzle flash peak P1. The effect of the clutter can be considered in more detail.
FIG. 4B exemplifies a signal SPD', a derivative of signal The clutter is composed of several components. The first
SPD of FIG. 4A. The derivation can be performed by means 65 component is the Sun radiation and reflections of Sun radia
of a differencing circuit or subtracting circuit (36 in FIG.3B). tion from various objects (e.g. from vegetation Such as grass
Additionally, before the derivation, a low pass filter can be and leaves); the maximum of this radiation is within the
US 8,809,787 B2
35 36
visible range. This Sun (or Solar) clutter is non-uniform and daylight or moonlight, it should be done so as to allow propa
impedes the muzzle flash detection. This is because its illu gation of Sufficiently wide wavelength range(s) to the photo
mination power and power variation may be similar to the detector. If the signal arriving to the photodetector is too
muzzle flash; for example its non-uniformities might cause small, the detection will be compromised by the internal (e.g.
signal glitches that might be wrongly interpreted as a muzzle 5 dark) noise of the detector and the shot noise of the signal. It
flash when the detector is moved. also should be noted that before arriving to the photodetector
The second clutter component (the so-called “blackbody the signal is attenuated by the atmospheric absorption and the
clutter' is a black- or gray-body radiation of the detector absorption of the filter. Thus, the exact edges of the filtered
environment (e.g. air, building walls, etc.). The detector envi wavelength range(s) can vary, so as to maximize the total
ronment has a relatively low temperature (when compared to 10 SNR ratio, in which both clutter and noise are taken into
the temperature of the Sun); it emits light, which wavelength acCOunt.
distribution has a maximum intensity in LWIR (e.g. 10 It should be noted that the control unit (or the layer-2
micron). It should be noted that the detector environment processing unit) can be configured to implement several pro
temperature is distributed non-uniformly; in some cases these
non-uniformities present a useful signal. 15 cessing tasks. First, in accordance with the above, it may be
The technique of the inventors can treat the above clutter configured to determine whether one or a group of the digital
components differently. According to the invention, the pho oranalog layer-1 output signals corresponds to a muzzle flash
todetector array may be of a type insensitive to the MWIR and signal. This may be done by processing the temporal and
LWIR ranges. Thus, the second clutter component (the black spatial features of the digital or analog signals, e.g. by com
body clutter) does not generate a significant signal at the 20 paring the intensity of the analog signals with a certain thresh
detector output. The solar clutter can be detrimental during old, or by comparing the received analog signals with various
the day or the night, if moonlight is present, but it can be muzzle flash signatures stored in a database in the memory
almost totally prevented by the narrowing of the imaged utility. For instance, if an analog processing unit differentiates
spectrum to the wavelength in the region(s) of low atmo the photodetector output or subtracts sequential readings of
spheric transmission, because at these wavelengths solar or 25 the photodetector output, the analog signals corresponding to
moon light reaching the earth Surface is attenuated to essen a muzzle flash event will have a “positive peak followed by
tially Zero intensity. negative peak” signature, e.g. a signature as in FIG. 4B. Also,
The case when both clutter components are eliminated the muzzle flash radiation can be focused on more than one
from the photodetector array signal may be preferred also pixel of the photodetector array (e.g. if the center of the
because it provides for significantly reducing the shot noise 30 muzzle flash focuses on a line separating the pixels). The
associated with the signal. The shot noise is proportional to control unit can take into account the division of the optical
the square of the total signal: thus, if clutter is not eliminated, signal between several adjacent pixels.
it causes a shot noise that could be comparable with the Second, the control unit may be configured to determine
muzzle flash portion of the signal. Moreover, this noise might the wavelength range(s) being optimal for detection (i.e.
occasionally produce glitches having temporal and spatial 35 maximizing the SNR ratio). It can facilitate operating the
features somewhat similar to those of the muzzle flash optical filter (e.g. filter 10) and/or the collecting optics so at to
because the shot noise would be non-uniform even if the achieve this high SNR and/or output these data for the opera
clutter causing it would be uniform. tor's review. Also, the control unit may estimate a distance
It should be noted that, as it follows from the blackbody between the detector and the muzzle flash, for example from
clutter wavelength distribution, the clutter and the shot noise 40 the detected intensity of the optical signal; as well as to
can be eliminated for the larger portions in the NIR rather than determine the type of weapon that was shot, etc.
the SWIR range. The control unit can be configured to tune the optical filter
Also, it should be noted, that the invented technique of So as to periodically change its passing band to define a time
maximizing the signal to noise ratio (SNR) is different from a multiplexing scheme for the muzzle flash detection in more
common technique. While the latter would suggest filtering a 45 than one wavelength range. For example, the photodetector
signal in a frequency range around the signals peak(s), the may be sensitive to two different subbands in the NIR/SWIR
technique of the inventors can utilize imaging in the wave ranges, or to a region of ultraviolet/visible range and a region
length regions being remote from the muzzle flash signal of NIR/SWIR range; etc. Thus various time-multiplexing
peak. schemes can be realized. It should be understood that the
The narrowing of the imaged spectrum to the band(s) of 50 relative intensities of signals of various wavelength bands
low atmospheric transmission is done by means of the optical carry additional information, for example about the distance
filter (which may be of any known type), which may be an to the muzzleflash, in particular due to the difference between
external filter or may be integrated with the photodetector atmospheric absorption coefficient being pertinent to the vari
array; e.g. by introducing a narrow band cavity within the ous wavelengths.
photodetector array. 55 It should also be noted, that the time-widths (durations) of
It should be understood that the use of the band filter is muzzle flash pulses corresponding to different bands are gen
beneficial when the solar clutter is the limiting factor. At erally different. For example, ultraviolet radiation is emitted
night, the Solar clutter is orders of magnitude lower and the as a result of the electron transitions between different
detector noise becomes the limiting factor. Thus, at night, the molecular levels during the chemical reaction of oxygen and
filter can be removed or not operated. This would improve the 60 burning powder, thus the time-width of the ultraviolet pulse is
gunshot probability of detection to false alarm count ratio, as small; the NIR/SWIR radiation pulse is a result of the thermal
well as would allow for night vision, when SWIR is used. radiation of the hottest gas; and the MWIR/LWIR is a result of
Hence, the filter, if being present in the device, is configured the thermal radiation of a cooler gas, thus the time-width of
in some of the device's preferred embodiments to be shiftable the MWIR/LWIR signal is longer. Within the time multiplex
between its operative and inoperative states. 65 ing scheme framework, the control unit changes the pass band
Although the narrowing of the imaged spectrum to the low of the optical filter with a rate of change selected to allow
atmospheric transmission regions can increase the SNR at detection of signals of the shortest time width.
US 8,809,787 B2
37 38
The control unit can be configured to operate digital or 7. The method of claim 1, wherein the imaging is substan
analog, typically parallel, layer-1 processing unit. For the tially within the range of low atmospheric light transmission
analog layer-1 processing unit this can be done, for example, at least partially including a trough extending from 1.80 um to
by applying Voltage to Switches in this unit, so as to selec 2.00 um.
tively direct the photodetector output to the circuits designed 5
8. The method of claim 1, wherein said filter passes less
for processing signals corresponding to input light of differ than 10% of energy of wavelengths being outside said spec
ent wavelength ranges, as in the time-multiplexing scheme. tral range of low atmospheric transmission and sensed by the
Referring to FIG. 6, there is schematically shown an PDA.
example of a weapons-firing detection system 200 of the
present invention. System 200 includes a muzzle flash detec 10 9. The method of claim 1, wherein said filter passes less
tor 100 (similar to either one of FIG. 2A or 2B), an acoustic than 2% of energy of wavelengths being outside said spectral
detector 70, and a control unit 210. Control unit 210 is con range of low atmospheric transmission and sensed by the
PDA.
figured similar to the above-described control unit and also is
adapted to receive and process the output of the acoustic 10. A device for use in detection and location of a muzzle
detector so as to compare the detection results of both detec 15 flash event, the device comprising an imager comprising a
tors. This way, most of the false alarms of each of the acoustic non cryogenically cooled Photo Detector Array (PDA) sen
and the muzzle flash detector can be avoided: control unit 210 sitive in at least a portion of the SWIR spectrum, and a filter
will generate an alarm only when each of the optical and the of electromagnetic radiation configured and operable for
acoustic detector identifies a gunshot, and a delay between selectively passing therethrough spectral bands correspond
these identifications is within some meaningful limit. The ing to relatively low transmission of the electromagnetic
delay between the identifications events at optical and acous radiation in atmosphere for said at least portion of the SWIR
tic detectors depends on the distance to the shooter and the spectrum.
speed of Sound at frequency sensed by the acoustic detector. 11. The device of claim 10, wherein said PDA comprises at
Therefore, this delay can be used for determination of the least 10,000 pixels.
distance to the shooter. 25
Those skilled in the art will readily appreciate that various 12. The device of claim 10, wherein the PDA has an inte
modifications and changes can be applied to the embodiments gration time shorter than 10 s.
of the invention as hereinbefore described without departing 13. The device of claim 12 wherein said PDA has a dead
from its scope defined in and by the appended claims. time shorter than a duration of the muzzle flash event.
The invention claimed is: 30 14. The device of claim 10, wherein the PDA has a sensi
1. A method for use in detecting and locating a muzzle flash tivity cut-offat a wavelength in between 1.4 um and 2.5um.
event, the method comprising imaging electromagnetic radia 15. The device of claim 10, wherein the device is adapted to
tion by an imager comprising a non cryogenically cooled sense electromagnetic radiation Substantially within a range
Photo Detector Array (PDA) sensitive in at least a portion of oflow atmospheric light transmission at least partially includ
a SWIR spectrum, wherein said electromagnetic radiation 35
ing the trough situated around 1.15um (micron).
propagating towards the PDA undergoes filtering for selec
tively passing towards the PDA the electromagnetic radiation 16. The device of claim 10, wherein the device is adapted to
of one or more spectral ranges of relatively low transmission sense electromagnetic radiation Substantially within a range
in atmosphere of said at least portion of the SWIR spectrum. oflow atmospheric light transmission at least partially includ
2. The method of claim 1, wherein said imager comprises at 40 ing a trough extending from 1.34 um to 1.50 um.
least 10,000 pixels. 17. The device of claim 10, wherein the device is adapted to
3. The method of claim 1, wherein an integration time of sense electromagnetic radiation Substantially within a range
the PDA is shorter than 10°s. oflow atmospheric light transmission at least partially includ
4. The method of claim 3, wherein the imaging recurring ing a trough extending from 1.80 um to 2.00 um.
with dead time shorter than a duration of the muzzle flash 45
18. The device of claim 10, wherein said filter passes less
event. than 10% of energy of wavelengths being outside said spec
5. The method of claim 1, wherein the imaging is substan tral range of low atmospheric transmission and sensed by the
tially within the range of low atmospheric light transmission PDA.
at least partially including the trough situated around 1.15um
(micron). 50 19. The device of claim 10, wherein said filter passes less
6. The method of claim 1, wherein the imaging is Substan than 2% of energy of wavelengths being outside said spectral
tially within the range of low atmospheric light transmission range of low atmospheric transmission and sensed by the
PDA.
at least partially including a trough extending from 1.34 um to
1.50 lum.

US8809787

Uploaded by

Copyright:

Available Formats

US8809787

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

US8809787

Uploaded by

Copyright:

Available Formats

US008809.

(12) United States Patent (10) Patent No.: US 8,809,787 B2

3,699,341 A 10/1972 Quillinan et al.

{{ISSNS - &£S. sGNALs *RS). -ie A {

S&N is gy Šiš, ri iF THE Sks:-83 : * {-g

~· ~); -)b–†6uis epboluad

You might also like