System and method for identifying location of gunfire from a moving object

Abstract

A system and method for identifying the location of gunfire from a moving object in which at least two spaced apart microphones are positioned on the moving object. The output from the microphones is decomposed into intrinsic mode functions by using empirical mode decomposition. A transient pulse is then identified in the intrinsic mode function representative of the gunfire. The location of the origin of the gunfire is then determined from the transient pulses through multilateration using time difference of arrival of the transient pulses by the microphones.

Claims

1. A method of identifying a location of gunfire from an object comprising the steps of: placing at least two spaced apart audio microphones on the object, each microphone producing an output signal representative of all received audio sound, decomposing said received audio sound into intrinsic mode functions by using empirical mode decomposition, identifying a transient pulse in said intrinsic mode functions for each microphone, locating the location of the gunfire from said transient pulses through multilateration using time difference of arrival of said transient pulses by said microphones, and displaying said location.

2. The method as defined in claim 1 wherein the object is a moving object.

3. The method as defined in claim 1 and further comprising the step of forming a matrix of said intrinsic mode functions prior to said identifying step.

4. The method as defined in claim 1 wherein said placing step comprises the step of placing at least three spaced apart audio microphones on the moving object, each microphone producing an output signal representative of all received audio sound.

5. The method as defined in claim 4 wherein said locating step further comprises the steps of: creating a hyperbola of the sound location for each microphone, calculating a location of an intersection of said hyperbolas.

6. The method as defined in claim 5 wherein said calculating step comprises spherical interpolation.

7. The method as defined in claim 1 wherein said identifying step further comprises the step of identifying a transient pulse in a high frequency intrinsic mode functions for each microphone.

8. A system to identify a location of gunfire from an object comprising: at least two spaced apart audio microphones positioned on the object, each microphone producing an output signal representative of all received audio sound, a processor programmed to decompose said received audio sound into intrinsic mode functions by using empirical mode decomposition, said processor programmed to identify a transient pulse in said intrinsic mode functions for each microphone, said processor programmed to locate the location of the gunfire from said transient pulses through multilateration using time difference of arrival of said transient pulses by said microphones, and a display which displays said location.

9. The system as defined in claim 8 wherein the object is a moving object.

10. The system as defined in claim 8 and further comprising the step of forming a matrix of said intrinsic mode functions prior to said identifying step.

11. The system as defined in claim 8 wherein said placing step comprises the step of placing at least three spaced apart audio microphones on the moving object, each microphone producing an output signal representative of all received audio sound.

12. The system as defined in claim 11 wherein said locating step further comprises the steps of: creating a hyperbola of the sound location for each microphone, calculating a location of an intersection of said hyperbolas.

13. The system as defined in claim 12 wherein said calculating step comprises spherical interpolation.

14. The system as defined in claim 8 wherein said identifying step further comprises the step of identifying a transient pulse in a high frequency intrinsic mode functions for each microphone.

Description

BRIEF DESCRIPTION OF THE DRAWING

(1) A better understanding of the present invention will be had upon reference to the following detailed description when read in conjunction with the accompanying drawing, wherein like reference characters refer to like parts throughout the several views, and in which:

(2) FIG. 1 is an elevational view illustrating a vehicle equipped with the system of the present invention;

(3) FIG. 2 is a flowchart illustrating the algorithm of the present invention;

(4) FIG. 3 is a graph illustrating sample data collected by a microphone with four transient pulses, representative signal corresponding to gunshot, at different speeds of vehicle;

(5) FIG. 4 is a graph illustrating the signal received by two microphones clearly showing multipath signals riding the noise immediately after the transient pulse;

(6) FIG. 5 is a graph of the decomposition of the audio signal into five intrinsic mode functions;

(7) FIG. 6 is a time graph of the identified transient pulse (gunshot) from one microphone and the multipath signals extracted (first intrinsic mode function) using EMD;

(8) FIG. 7 is time graph illustrating the product of first intrinsic mode functions of all microphones and a box indicating the extent the multipath signals persist; and

(9) FIG. 8 illustrating the TDOA.

DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT OF THE PRESENT INVENTION

(10) With reference first to FIG. 1, a vehicle 10 is shown. The vehicle 10 may be any sort of vehicle, such as a jeep, armored vehicle, tank, and/or the like, and preferably is of the type of vehicle used by the military. The vehicle 10 further has at least two, and preferably three or more, audio microphones 12 that are attached to the outside of the vehicle 10. Each microphone produces an output signal representative of all received audio sound over the audio spectrum of the microphone.

(11) The outputs from each microphone 12 are coupled as input signals to an interface 14, such as an analog/digital (A/D) converter 14. In the well-known fashion, the A/D converter 14 converts the received analog audio signals from the microphones 12 and converts these audio signals into a digital output from the A/D converter 14. That output 16 is coupled as an input signal to a processor 18. In a fashion to be subsequently described in greater detail, the processor 18 is programmed to not only identify distant gunfire using the audio inputs from the microphones 12, but to also locate the source of the gunfire relative to the vehicle 10. The processor 18 then generates an output signal to a display device 20 which is visible to personnel in the vehicle 10.

(12) With reference now to FIG. 2, the processor 18 is programmed to receive the audio signals from the microphones 12 at step 40. Although any conventional method may be utilized to receive the now digitized audio signals from the microphones 12 and to store those signals for access by the processor 18, preferably, a WaveBook data acquisition system is used to acquire the audio signals. An exemplary sampling rate of 4000 samples per second is sufficient to provide rapid localization of the gunfire using muzzle blast signals and display that localization on the display 20. For supersonic bullets, one may want to capture the shockwave signals for localizing the gunfire. In which case the sampling rate in excess of 20 K samples per second will be used to capture the shockwave. After the digitized audio signals are received at step 40, step 40 proceeds to step 42. FIG. 3 shows the signals captured by one of the microphones mounted on a vehicle traveling at 25-30 miles per hour. The first transient pulse shown at about 4 sec is clearly above noise as the distance between the vehicle and the source of gunfire (or loud noise generator) is only 60 meters. Other three pulses are buried in noise. Enlarged portions of the signals at 20 and 50 sec marks are shown also in FIG. 3.

(13) The gunshot localization requires the time difference of arrival (TDOA) of muzzle blast signal (transient pulse) at multiple microphones. In order to determine the TDOAs between pairs of microphones, the muzzle blast signals should be detected. However, detection of muzzle blast signals in the presence of platform and flow noise is difficult. But if we know where they might be, detection would be easy. Each muzzle blast signal is followed by the multipath signals (that is, muzzle blast signals bounced off by various objects). The signal received by each microphone can be represented as:

(14) s(t) captured by a microphone due to gunfire as sum of multiple components:

(15) $\begin{array}{cc}s\left(t\right)=A_{0}x\left(t\right)+\underset{i=1}{\overset{M}{.Math.}}{A}_{i}x\left(t+{}_i\right)+n\left(t\right)& \left(1\right)\end{array}$
where x(t) is the muzzle blast signal emitted by the gun and x(+.sub.i) is the i.sup.th multipath signal received by the microphone, A.sub.i is its amplitude and n(t) is the noise. The first component A.sub.0x(t) in Equation 1 is the direct path signal received by the microphone and its amplitude A.sub.0 is predominant compared to multipath signal amplitudes A.sub.i. The multipath signals arrive after the direct path signal and superimpose on the noise. This is seen in FIG. 4 clearly as additional fluctuations in noise immediately after the occurrence of transient pulse. This feature is used to localize the occurrence of the transient event, i.e., the time when the additional fluctuations in signal occur is identified. In order to detect these additional fluctuations due to multipath, we decompose the received signal into its intrinsic mode functions using empirical mode decomposition (EMD).

(16) At step 41 in FIG. 2 the processor 18 decomposes the received audio signals into intrinsic mode functions utilizing empirical decomposition which is based on the direct extraction of the energy associated with the various intrinsic time scales. Furthermore, the intrinsic mode functions have well behaved Hilbert transforms from which the instantaneous frequencies can be calculated. FIG. 5 shows the decomposition of a signal at the 54 sec mark shown in the upper half of FIG. 4 into five intrinsic mode functions. In FIG. 6, the signal at the 54 sec mark and its first intrinsic mode function is shown. From this figure, we can see the multipath signals are extracted clearly by the first intrinsic mode function with higher amplitudes than the noise prior to the occurrence of transient pulse.

(17) At step 44, the processor 18 identifies the location where the multipath signals are present due to muzzle blast travelling different paths. The algorithm used for determining the location is given below: 1. Let s.sub.i be the signal corresponding to microphone i{1, 2, . . . , n}, where n denotes the number of microphones. 2. Let e.sub.i=emd(s.sub.i), where emd is the empirical mode decomposition of signal s.sub.i and e.sub.i=[e.sub.i.sup.1, e.sub.i.sup.2, . . . , e.sub.i.sup.m].sup.r is a matrix of m intrinsic mode functions. 3. Compute g=.sub.j=1.sup.ne.sub.i.sup.1. 4. Set G(t)=1 if g(t) else G.sub.i(t)=0, where is some threshold and G(t) is set to 1 only if the time interval t>0.1 sec.

(18) With reference to FIG. 7, the product of first intrinsic mode functions of all microphones g and the output of the algorithm G above is illustrated in which a pulse 28 is generated. The pulse 28 is shown in FIG. 7 as a function of time on the X axis gives the location of the multipath signals. The transient pulse (muzzle blast) signal occurred just prior to the pulse G.

(19) After identification of the location of the transient pulses at step 42, step 42 proceeds to step 43 where the time difference of arrival between each pair of microphones is estimated using the technique called generalized cross correlation (GCC) presented in C. H. Knapp and C. Carter, The generalized correlation method for estimation of time delay, IEEE Trans. on Acoustics, Speech and Signal Processing, Vol. ASSP-24, No. 4, August 1976, pp. 320-327. FIG. 8 shows an exemplary TDOA between two microphones. Once the TDOAs for every pair of microphones are determined the location of the gunfire is identified through multilateration using time difference of arrival (TDOA) of the transient pulses by the microphones 12. Such localization is straightforward and may be performed by the following algorithm.

(20) Let the instant the gun's muzzle blast is emitted to be to and the time the signal arrived at the microphone S.sub.i to be t.sub.i, then the distance the sound traveled is
r.sub.i=(t.sub.it.sub.0)c,(2)
where c denotes the speed of sound. Using Equation 2, the TDOA between two microphones, S.sub.i and S.sub.j, is given by
t.sub.ij=(t.sub.it.sub.0)(t.sub.1t.sub.0)=t.sub.it.sub.j(3)
and the difference in the distances is
r.sub.ij=(t.sub.it.sub.0)c(t.sub.jt.sub.0)c=(t.sub.it.sub.j)c=r.sub.ir.sub.j.(4)
The signal source must lie on the locus, which keeps the difference r.sub.ij constant. The locus defines a hyperbola. If there are at least three microphones, the intersection of the hyperbolas gives the location of the signal source, that is, location of the gun.

(21) The following approach gives the procedure to find the point of intersection of the hyperbolas. Let S=(x,y) denote the location of the sound source to be estimated; the locations of the microphones. S.sub.i=(x.sub.i, y.sub.i) are known. Without loss of generality, the location of S.sub.l is set at (0, 0) (just subtract the coordinates of all microphones with the coordinates of S.sub.l to make S.sub.l(0, 0)). Now, from Equation 4
r.sub.il=r.sub.jr.sub.l
or r.sub.il+r.sub.l=r.sub.i={square root over ((x.sub.ix).sup.2+(y.sub.iy).sup.2)}
or (r.sub.il+r.sub.l).sup.2=K.sub.i2x.sub.ix2y.sub.iy+x.sup.2+y.sup.2=K.sub.i2x.sub.ix2y.sub.iy+r.sub.l.sup.2(5)
where K.sub.i=x.sub.i.sup.2+y.sub.i.sup.2 and r.sub.l.sup.2=x.sup.2+y.sup.2. Equation 4 can be rewritten as
x.sub.ix+y.sub.iy=r.sub.nr.sub.1+(K.sub.ir.sub.n.sup.2).(6)
Explicitly writing for all microphones, the above equation becomes

(22) $\begin{array}{cc}\left[\begin{array}{cc}{x}_2& y_2\\ x_3& y_3\\ \mathrm{.Math.}& \mathrm{.Math.}\\ x_n& y_n\end{array}\right]\left[\begin{array}{c}x\\ y\end{array}\right]=r_1\left[\begin{array}{c}-r_{21}\\ -r_{31}\\ \mathrm{.Math.}\\ -r_{n1}\end{array}\right]+\frac{1}{2}\left[\begin{array}{c}{K}_2-r_{21}^2\\ K_3-r_{31}^2\\ \\ K_n-r_{n1}^2\end{array}\right].& \left(7\right)\end{array}$
This is in the form of a linear equation
HX=r.sub.1G+D,(8)
where X=[x y].sup.T. The least-squares solution in terms of r.sub.1 yields
{circumflex over (X)}=(H.sup.TH).sup.1H.sup.T(r.sub.1G+D).(9)
Substituting this intermediate result into r.sub.1.sup.2=x.sup.2+y.sup.2 leads to a quadratic equation in r.sub.1. Solving for r.sub.1 and substituting the positive root back into Equation 9 yields the final solution for X. This method is called the spherical interpolation.

(23) Once the location of the gunfire has been identified, the processor 18 (FIG. 1) generates the appropriate output signal to the display 20. The war-fighters can then take the appropriate action.

(24) From the foregoing, it can be seen that the present invention provides a system and method for identifying the location of gunfire from an object, such as a moving object. Unlike the previously known systems for identifying the location of gunfire, the present invention is able to extract transient pulses representative of the gunfire even when the transient pulse is buried in noise.

(25) Having described my invention, however, many modifications thereto will become apparent to those skilled in the art to which it pertains without deviation from the spirit of the invention as defined by the scope of the appended claims.