WO2006069596A1 - Method for determining the position of impacts - Google Patents

Abstract

A method for determining the position of impacts on an object comprising two acoustic sensors, and N active areas of said object, comprises the steps of: (a) receiving two acoustic signals S1(t) and S2(t); (b) calculating a sample signature function SIGS (ω) = S1 (ω)- S2 (ω)*, where S1(ω) and S2 (ω) are the respective Fourier transforms of S1(t) and S2(t) , (c) comparing SIGS(ω) with N predetermined reference signature functions SIGR,(ω) corresponding to the predetermined area j for j from 1 to N; (d) determining the active area in which the impact occurred, on the basis of the comparison of step (c).

Description

METHOD FOR DETERMINING THE POSITION OF IMPACTS

FIELD OF THE INVENTION

The present invention relates to a method for determining the position of impacts, and a device using this method.

BACKGROUND OF THE INVENTION

A known method for determining the position of impacts on an object is suggested by Patent FR 2 841 022. According to this document, the object may comprise two acoustic sensors and N predetermined active areas of said object, whereby N is an integer at least equal to 1. In order to determine the active area in which an impact occurred, acoustic signals received by the acoustic sensors are compared with N acoustic signals, recorded in a database, and each corresponding with one of the N predetermined areas. Consequently, it leads to 2N comparison calculations to complete the task, when two sensors are used. An object of the present invention is to provide a method for determining an impact position, whose calculation velocity is improved, and which does not require a high computing power.

SUMMARY OF THE INVENTION The invention thus proposes a method for determining the position of impacts on an object, said object comprising: two acoustic sensors;

N predetermined active areas, whereby N is an integer at least equal to 1; said method comprising the steps of:

(a) receiving two acoustic signals S₁(O ^and $ (t) originating respectively from said acoustic sensors and generated by an impact received on said object; (b) calculating a sample signature function:

where Sι(^ω) ^and SiM ^{are the} respective Fourier transforms of $_{(t) and

^anc* where * is the complex conjugate operator; (c) comparing SIGv M with N predetermined reference

signature functions SIGR- M ^e^ch corresponding to a

predetermined active area j, for j from 1 to N; (d) determining an active area in which the impact occurred, on the basis of the comparison of step (c) . In various embodiments of the method according to the invention, at least one of the following characteristics may be used: each reference signature function equals:

SIG_n M = R₁MR₂M where R_λJ(fi>) and J^₂.(ω) are Fourier transforms of acoustic signals ^₁.(t) and y₂.{t) received by each of the respective acoustic sensors when an impact occurs on the predetermined area j ; step (c) comprises the calculation of a similarity estimator a_} representing a function of the

^phas^{e φ(}CORj(ω))o^f

^(ω)* _'' step (c) comprises the calculation of a function δ_j(a>) for j from 1 to N, wherein δ_j{ώ)=ε_k, if φ(CORj(ω)) belongs to J_k,

where ε_k is a predetermined value and J_k a corresponding angular interval for k from 1 to n, where n is an integer greater than 1; the values ε_t are not greater than 1; in the method: if |φ(CORj(ω))| < al, then δ_j(ω)= E₁,

if n is greater than 2 and a_k-i < |φ(COR_j(ω))| ≤ a_k, then S_j(ω)= ε_k, for k=2... n-1,

if |φ(CORj(ω))| > an-!, then δ_j(ω)= ε_n, wherein a_k is increasing with k and ε_k is decreasing with k; - in the method: if Re(CORj(ω))> | Im(CORj(ω)) | / tan(ai) , then δ_j{ω)= ε_lf

if n is greater than 2 and | Im(COR_j(ω)) | /tan (a_k-i) > Re(CORj(ω))> | Im(COR_|(ω)) I /tan (a_k) , then δ_j{ω)= ε_k, for k=2... n-1,

else, δ_j(ω)= ε_n, and wherein Re (COR_j(ω)) is the real part of COR_j(ω) and Im(CORj(ω)) is the imaginary part of COR_j(ω); - in the method: if Y_j (ω) ≥ βj(ω)/tan (ai) , then δ_j(ω)= E₁ = 1;

if n is greater than 2 and β_j (ω) /tan (a_k-i) > Y_j (ω) ≥ β_j (ω) /tan (a_k) , then δ_j(ω)= ε_k, for k = 2, ... n-1;

else, δ_j(ω)= ε_n, and wherein: β_j(ω) =|l+{A_j(ω)/B_j(ω) } . { (D(ω) /C (ω) } I ,

Y_j (ω)=signBj (ω) .signC(ω) . [ {A_j (ω) /B_j (ω) }-{D(ω)/C(ω) }] , signBj (CO) = 1 if Bj (ω) is positive and = -1 if B_j (ω) is negative, signC(ω)= 1 if C(ω) is positive and = -1 if C(ω) is negative, A_j(ω) and B_j (ω) are respectively the real part and the imaginary part of each reference signature function

SIG_Rj ^(«).

C(ω) and D(ω) are respectively the real part and the imaginary part of the complex conjugate sample signature function SIGs(Cύ)*; εi = 1 and ε_n = 0; n is greater than 2 and εk = cos (a_k-i) , for k = 2, ... n-1.

where B is a freguency interval and K a constant ;

B = [ω_min, ω_max] and K is proportional to 1/ (ω_max . k>min) / the active area j₀ in which the impact occurred is determined such that a_js> is the greatest similarity estimator among the N calculated similarity estimators a_} ; it is concluded that the impact occurred in the area J₀ only if a_h is considered as valid in step (d) ; α_y is considered as valid if a_} is greater than a predetermined threshold of confidence; - it is concluded that there is no impact if a_Jo is not considered as valid in step (d) ; step (a) begins if the acoustic signals ^₁(J) and g (t) are above a predetermined trigger threshold; the acoustic signals g^t) and £₂(0 in step (a) are delayed relative to real audio signals; the Fourier transform is a Fast Fourier transform. Besides, another object of the invention is a device for determining the position of impacts on an object, by comprising: two acoustic sensors adapted to be borne by said object for receiving acoustic signals 5,(0 ^and £₂(0 generated by an impact on said object; memory means comprising N reference signature functions, corresponding to N predetermined active areas of said object, whereby N is an integer at least equal to

1; - calculation means for calculating a sample signature function

SIG_s ^(»)=5,^(«)■S₂ ⁽">^)* _' where SS^ω^ ^anc* SA^ω) ^{are tne} respective Fourier transforms of ^₁(O and $ (t) , and where * is the complex conjugate operator; comparison means for comparing SIGv (^ω) with N

predetermined reference signature functions SIG_D (^ω) ^^or J

from 1 to reprocessing means for determining an active area (Ia) in which the impact occurred, on the basis of results given by said the comparison means.

In various embodiments of the device according to the invention, one may use one and/or other of the following features: - each reference signature function equals:

where R_r(^ω) and R₂X^ω) ^are Fourier transforms of acoustic signals ^₁.(0 and y₂.(t) received by each of the respective acoustic sensors when an impact occurs on the predetermined area j ; said comparison means are adapted to calculate of a similarity estimator α_y representing a function of the phase φ(CORj(ω)) of CORj(ω)=SIG_R WSIG₈ W_' said comparison means are adapted to calculate a function δ_j(ω) for j from 1 to N, wherein δ_j(ώ)=ε_k, if φ(COR_j(ω)) belongs to J_k, where ε_k is a predetermined value and J a corresponding angular interval for k from 1 to n, where n is an integer greater than 1; - the values £", are not greater than 1; said comparison means are adapted so that: if I φ (CORj(CO))I < al, then δ_j(ω)= ε_lt

if n is greater than 2 and a_k-i < |φ(COR_j(ω))| ≤ a_k, then δ_j{ω)= ε_k, for k=2... n-1,

if |φ(COR_j(ω))| > a_n.i, then δ_j(ω)= ε_n, wherein a_k is increasing with k and ε_k is decreasing with k; said comparison means are adapted so that: if Re(CORj(ω))> | Im(CORj(CQ)) | / tan(aχ), then δ_j(ω)= ε_Xr if n is greater than 2 and | Im(COR_j(ω)) |/tan(a_k-i) > Re(CORj(ω))≥ | Im(CORj(ω)) I /tan (a_k) , then δ_}{ω)= ε_k, for k=2... n-1,

else, δ_j(ω)= ε_n, and wherein Re(CORj(ω)) is the real part of COR_j(ω) and Im(COR_j(ω)) is the imaginary part of COR_j(ω); said comparison means are adapted so that: if Yj (Co) > βj(ω)/tan (ai) , then δ_j{ω)= E₁ = 1;

if n is greater than 2 and β_j (co) /tan (a_k_i) > Y_j (ω) ≥ βj (ω) /tan (ak) , then δ_j{ω)= ε_k, for k = 2, ... n-1;

else, δ_j(ω)= ε_n, and wherein: βj (CO) =I l+{Aj(ω) /Bj (ω) }. { (D(CD) /C(ω) } I,

Y_j(co)=signBj (ω) .signC(ω) . [{A_j(co) /B_j (ω) }-{D(ω) /C (ω) } ] , signBj (ω) = 1 if Bj (co) is positive and = -1 if B_j (co) is negative, signC(co) = 1 if C(ω) is positive and = -1 if C(ω) is negative,

Aj (co) and Bj (ω) are respectively the real part and the imaginary part of each reference signature function

SIGφ), C(ω) and D(ω)are respectively the real part and the imaginary part of the complex conjugate sample signature function SIGsfco)*; εi = 1 and ε_n = 0; n is greater than 2 and ε_k = cos(a_k-i), for k = 2, ... n-1; said comparison means are adapted so that: a_j =K. [δ_j(ω).dω, where B is a frequency interval and K a constant; the processing means are adapted to determine the active area j₀ in which the impact occurred such that a_} is the greatest similarity estimator among the N calculated similarity estimators α. ; the processing means are adapted to determine that the impact occurred in the area j₀ only if a_} is considered as valid; the processing means are adapted to determine that a_jo is valid if a_Jo is greater than a predetermined threshold of confidence; the processing means are adapted to determine that there is no impact if a_jo is not considered as valid.

This method and this device permit to reduce the number of comparisons to N, as the sampled acoustic signals are no longer compared with each reference acoustic signal, but with a signature function for each pair of reference acoustic signals. Furthermore, the calculation of the signature functions does not require a high calculation power. Moreover, as explained below, the phase of a signature function does not depend on the excitation waveform, but only on the place in which the impact occurred. Consequently, by studying only the phase of the signature function, the active area in which the impact occurred may be determined. BRIEF DESCRIPTION THE DRAWINGS Other features and advantages of the invention will appear from the following description of three embodiments of the invention, given by way of non-limiting example, with regard to the appended drawings. In the drawings: - figure 1 is a schematic view of an example of device according to the invention; figure 2 is a flow chart describing a method according to a first embodiment of the invention; figures 3A and 3B are graphics illustrating the calculation of a similarity estimator according to the first embodiment of the invention; figures 4A and 4B are graphics illustrating the calculation of a similarity estimator according to a second embodiment of the invention; and - figure 5 is a flow chart describing a method according to a third embodiment of the invention.

MORE DETAILED DESCRIPTION

As illustrated by Figure 1, the present invention may be carried out for instance by a device comprising an object 1 and two acoustic sensors SENSl and SENS2 borne by said object 1.

This object 1 may be for instance a table, a door, a wall, a screen or other things, and is made for instance of wood, metal, cement, glass or other materials. The acoustic sensors SENSl and SENS2 may be for example piezoelectric sensors or any other sensor able to sample a vibration transmitted in the object, such as capacitive sensors, magnetostrictive sensors, electromagnetic sensors, acoustic velocimeters, optical sensors (laser interferometers, laser vibrometers) , etc.

The output of the sensors SENSl, SENS2 may be connected respectively to amplifiers 3 and 4, the output of which is respectively connected to filters (F) 5, 6. The bandwidth of each filter 5 and 6 may be in the range of 300 Hz to 20000 Hz. The output of the filters 5, 6 is connected to a multiplexer 7, the output of which is connected to an analog to digital converter (ADC) 8. The output of the analog to digital converter 8 is connected to a processing unit (PU) 8 such as a microprocessor, a microcontroller, a DSP (digital signal processor) , a programmable logical array (CPLD, FGPA), etc. The processing unit 9 may be connected to a RAM 9a and to a ROM 10.

The ROM 10, or any other memory communicating with the processing unit 9, contains a database of N reference signature functions £JQ_R. (ω) , for j=l...N, corresponding to N

predetermined active areas Ia of the object 1, as will be described hereafter. Each of these active areas may correspond for instance to a particular information, and may be materialized or not by a marking or other indication on the object. ^' The ROM 10 may be a hard disk, but also an EPROM, or an EEPROM, or likewise. Using an EEPROM can permit to reconfigure easily the device, for example for other applications, by recording new active areas Ia.

The reference signature functions may be for instance calculated during a learning step, e.g. before the device is used in normal use. The reference signature functions may be calculated for each single device 1-10, or these reference signature functions may be calculated only once for one device 1-10 and then used for all identical devices during normal use.

During the learning step, one generates an impact on each predetermined active area Ia of the object 1, which is sensed by sensors SENSl, SENS2. For each impact on an active area j (j = 1 ... N), the sensors SENSl, SENS2 generate acoustic signals which are respectively amplified by amplifiers 3, 4, filtered by filters 5, 6, multiplexed by multiplexer 7 and sampled by analog to digital converter 8. Analog to digital converter 8 thus outputs two reference signals rχ_j(t), r_2j(t) originating respectively from the two sensors SENSl, SENS2.

Respective Fourier transforms Rij (ω) , R₂j (ω) of the reference acoustic signals ri_j (t) , r_2j (t) are then computed. The Fourier transform R^_j (ω) of each reference acoustic signal equals:

where Ci (ω) is the Fourier transform of the impulse response of sensor i (i = 1, 2), H_RIJ (ω) is a propagation function, dependant on an acoustic wave propagation path in the object between active area j and sensor i, and E_Rj (ω) is the Fourier transform of the impact waveform on active area j .

Then, a reference signature function is calculated: SIG R_j («) = R_1J W-R_2J («)^* . ^{( 2 )} where * is the complex conjugate operator.

When using equation (1) to develop equation (2), one obtains:

SIGR_j (*>)= C_x(ω)H_RXj(ω)Ε_Rj(ω).C₂(_<ω)\H_R2j{ω)\E_Rj(ωy. (3)

Ey(CO) .E_RJ(ω)^* =

, so that equation (3) is equivalent to equation (4) : SIGBJ (^)=C_i(ω)C₂(ωy.H_mj(ω).H_R2J(ω)'\E_RJ(ω)\² (4)

During normal use of the device, when an impact is generated on the object 1, e.g. when a user hits or touches the object 1 with a finger or with another object

(stylus, pen or else) the sensors SENSl, SENS2 receive acoustic signals. These acoustic signals are amplified by amplifiers 3 and 4, filtered by filters 5 and 6, multiplexed by multiplexer 7 and then sampled by analog to digital converter 8. The sampled signals 5,.(O originating respectively from the sensors i (i=l, 2 in the present example) are then processed by processing unit 9, for instance according to a first embodiment of the method of the invention as shown in the flow chart of figure 2. This method may be carried out by a program ran on processing unit 9, comprising for instance calculation means SlOl - S103, comparison means S104 - SIlO and processing means Sill.

In this embodiment, respective Fourier transforms S_\(^ω) ^and 5^* ₂^ ^{of the two sam}Pl^ed acoustic signals ^₁(O and 5₂(0 ^are computed (preferably only for positive frequencies ω) by calculation means in steps SlOl and S102. These Fourier transforms may be Fast Fourier Transforms (FFT) , permitting thus to obtain quick results, without a high calculation power.

The Fourier transform $.(#>) ⁰^ each acoustic signal equals:

where C₁(O)) is the Fourier transform of the impulse response of sensor i, H_Si(ω) is a propagation function, dependant on an acoustic wave propagation path in the object between the location of the impact and sensor i, and E_s(ώ) is the Fourier transform of the impact waveform.

At step S103, the calculating means compute a sample signature function: where * is the complex conjugate operator.

When using equation (5) to develop equation (6), one obtains:

SIG s («) = ^C> WJI (°>)-^Es WC₂ (ω)^' -H₃₂ (ω)' .E₅ {ω)\ ^[ I ⁾

E_s (ω) .E_s(ω)^* =

, so that :

SIG₈ {ω)

( 8 )

After initializing comparison means at step S104, for each predetermined area j of the object, an intermediary- estimator (correlation function) is calculated at step S105:

When using equations (4) and (8) to develop equation (9) one obtains equation (10) :

CORj(ω) = |Ci(ω)|².|C₂(ω)|². |E_Rj(ω)|².|Es(ω)|².H_R1j(_ω).H_R2j(ω)^*.Hsi(ω)^*.Hs₂(ω) (10)

Since |Ci(ω)|², |C₂(ω)|², |E_Rj(ω)|² and |E_s(ω)|² are square moduluses whose phase equals zero, the phase φ(COR_j(ω)) of the intermediary estimator COR_j(ω) does not depend on the impulse responses C-ι(ω), C₂(Co)Of the sensors 3, 4 and on the impact waveforms E_Rj(ω), Es(ω). This phase φ(COR_j(ω)) depends only on the phase of H_Ri_j(ω).H_R2j(ω) .Hsi(ω) .Hs₂(ω), i.e. on the acoustic wave propagation paths in the object during the learning step and the .

The method of the present invention is based on the observation that: if the impact which generated the sampled signals Si(t), S₂ (t) was not on active area j, then the phase of CORj(ω) is variable and different from 0, whereas if the impact which generated the sampled signals Si(t), S₂ (t) was on active area j, then the phase of COR_j(ω) is equal to 0, since:

COR_j(ω) = |C₁(co)|².|C₂(co)|². |E_RJ(CO)|².|ES(CO)|².|H₁J(CO)|².|H₂J(CD)|² (11).

Thus, it is possible to determine whether an impact on the object 1 was made on any active area Ia, and if so, to determine on which active area.

Since the method of the invention is based on direct or indirect determination of the phase of COR_j(ω), it should be noted that the accuracy of this method is not altered by the use of different types of impact or by the use of sensors 3, 4 of different characteristics, because this phase is in any case independent from the impact waveforms used during the learning step and during normal use, and because this phase is also independent from the responses of the sensors SENSl, SENS2. In steps S106 and S107, the phase φ(COR_j(ω)) is examined, to estimate how ^■ much SIG_c (^ω) matches with

SJζ_jn- (of) . To this end, the comparison means may calculate

a function δ_j(ώ) in Step 106, as shown figures 3A and 3B. This function δ_j{ω) may be calculated as follows: - if the phase φ(CORj(ω)) belongs to, for example, an angular interval J =[bl;al] , with ai>0 and bi<0, then δ_j(ω)= E₁ = 1; else, δ_j(ω)= ε₂ = 0.

It should be noted that, in step S106, it is not necessary to compute φ(COR_j(ω)). For instance, if bi = -al, the function δj (ω) may as well be calculated as follows: if Re(CORj(ω)) > I Im(COR_j(ω)) | / tan(al), then δ_j(ω)= εi = 1; else, δ_j(ώ)= ε₂ = 0. In step S107, the function δ_j{ώ) is integrated on a study frequency interval B = [ω_mi_n,ω_max], in order to obtain a similarity estimator a_j =K. [δ_j(ω).dω, where K is for example a constant of normalization, which equals to (or is proportional to) the inverse of the length of the interval B: K = l/(ω_max - ω_mi_n) . This estimator CXJ is simple to compute, does not require a high calculation power and correctly represents the similarity of the sampled signals Si(t), S₂ (t) with the reference signals ri(t), r₂(t) .

Once the N similarity estimators a_t are calculated and stored in a memory, the maximal similarity estimator a. is determined by comparison means in step SIlO. This can be carried out with a simple sequence of comparisons. In a variant, the maximum similarity estimator a_jf) can be determined in each iteration of the comparison means when computing the similarity estimators ct_j .

In order to eventually determine the active area in which an impact occurred, the greatest similarity estimator a_jo has to be greater than a predetermined threshold of confidence VAL in step Sill. For example, a has to be greater than VAL = 0.5 to be recognized as valid. If CC_j0 is greater than this threshold of confidence

VAL, the active area j₀ is determined by processing means as the active area in which an impact occurred. Then, an information associated to this active area may be transmitted to a software, or an action can be launched by the processing means 9, etc. Else, the impact is considered as interference or as perturbation. In a second embodiment of the invention, several angular intervals J. may be used in step 106, the other steps remaining the same as in the first embodiment. For example, Figures 4A and 4B illustrate such a construction in the example of 4 possible values of ot_j according to the phase φ(CORj(ω)). There, the phase φ(COR_j(ω)) is studied as follows: if the phase φ(COR_j(ω)) belongs to the interval Z₁ =[^₁Ja₁], with ai>0 and bi<0, then , then δ_j(ω)= εi = 1; if the phase φ(COR_j(ω)) belongs to the interval I₂

, with ε₂ < 1; if the phase φ(COR_j(co)) belongs to the interval J₃ =[b₃;b₂[u]a₂;a₃], with 180°>a₃>a₂ and -180°<b₃<b₂ (preferably with 90°>a₃>a₂ and -90°<b₃<b₂), then S_j(ω) =£₃, with ε₃ < ε₂ ; else, δ_j(ώ)= ε₄ = 0. As in the first embodiment of the invention, it should be noted that, in step S106, it is not necessary to compute φ(CORj(ω)). For instance, if bi = -ai, b₂ = -a₂, and b₃ = -a₃, then the function δj (ω) may as well be calculated as follows: - if Re(COR_j(ω)) ≥ | Im(COR_j(ω)) | / tan (ai) , then δ_j(ω)= E₁ = 1; if |Im(CORj(ω)) I / tan (ai) > Re(COR_j(ω)) ≥ I Im(COR_j(ω)) I / tan (a₂) , then δ_j(ω)= ε₂; if |Im(CORj(ω)) I / tan (a₂) > Re(COR_j(ω)) ≥ I Im(COR_j(Co)) I / tan (a₃) , then δ_j(ω)= ε₃; else, δ_j{ώ)=Q.

Re(COR_j(ω)) and Im(COR_j(ω)) are respectively the real part and the imaginary part of the complex number COR_j(ω).

For instance, ε₂ and ε₃ could respectively equal cos (ax) and cos(a₂) . More generally, the phase of COR_j(ω) could be compared to a plurality of n angular thresholds 0<ai<a2<...<a_n as follows: if Re(COR_j(ω)) ≥ |Im(COR_j(ω)) | / tan (ai) , then S_j(ω)= ε_x = 1; if I Im(COR_j(co)) | / tan (ai) > Re(COR_j(co)) ≥ |Im(COR_j(ω)) I / tan (a₂) , then δ_j(ω)= ε₂; if |Im(COR_j(ω)) | / tan (a₂) > Re(CORj(ω)) > |Im(COR_j(ω)) I / tan (a₃) , then δ_j{ω)= ε₃;

if I Im(COR_j(ω)) I / tan (a_k-i) > Re(COR_j(ω)) ≥ |Im(COR_j(ω)) I / tan (a_k) (12)

, then δ_j(ω)= ε_k;

- if I Im(COR_j(Co)) | / tan (a_n_₂) > Re(CORj(ω)) ≥ |Im(COR_j(ω)) I / tan (a_n_i) , then δ.(ω)= ε_n-i; else, δ_j(ω)= ε_n = 0.

The values ε_k are such that εi>ε₂> ... >ε_n>0, and these values may for instance equal: ε_k = cos (a_k-i) , for k = 2, ... n and ε_n+i = 0.

In this second embodiment the other steps of the method can be identical to those described above with regards to the first embodiment.

In a third embodiment of the invention, as illustrated for instance in figure 5, the calculation of the above correlation function COR_j (ω) (j = 1, ... N) is avoided.

As a matter of fact, SIG_Rj (ω) = A_j (ω) +j_B_j (ω) (A_j and B_j are respectively the real and imaginary parts of SIG_RJ) , and SIG₃(ω) * = C(ω)+j_D(ω) (C and D are respectively the real and imaginary parts of SIG₃) , so that the above equation (12) can be written as follows:

|B_j (ω) C(ω) +A_j (ω) D(ω) I /tan(a_k_i) > Aj (ω) C (ω) -Bj (ω) D(ω) > |B_j (ω) C (ω) +A_j (to) D(ω) |/tan(a_k) (13)

<=>

|Bj(ω) I . |C(ω) I . |l+{Aj(ω)/Bj(ω) }.{ (D(ω)/C(ω) } |/tan(a_k_i) >

|Bj(ω) I . |C(ω) I . |l+{Aj(ω)/Bj(ω) }.{ (D(ω)/C(ω) } |/tan(a_k) (14)

β_j (ω) /tan(a_k_i) > Yj (ω) > βj (ω) /tan (a_k) (15)

wherein: β_j(ω) = I 1+(A_j(CO) /B_j (ω) }.{ (D(CO) /C(CO) } |

Y_j (co)=signBj (ω) .signC(ω) . [ {Aj (ω) /Bj (ω) }-{D(ω) /C (co) }] ; signB_j (ω) = 1 if Bj (ω) is positive and = -1 if B_j (ω) is negative; signC(ω)= 1 if C(co) is positive and = -1 if C(ω) is negative.

Therefore, this third embodiment, instead of calculating the correlation functions COR_j (ω) , the above functions βj (co) and Yj (ω) are calculated in step S105. It should be noted that this calculation is particularly simple, and only requires: that signBj (ω) and the ratio A_j (ω) /B_j (ω) be stored in advance in memory 10 for j=l, ... N (i.e. one real number + 1 bit, whereas the complete calculation of the

COR_j (ω) implied to store the complete reference signature functions (a complex number , i.e. the equivalent of two real numbers) ; that signC(ω) and the ratio D(ω)/C(ω) be stored at the beginning of step S105, after which only a very limited number of operations remain to be done at step S105 to calculate βj (ω) and Y_j (ω) . Then, at step S106, the phase of COR_j(ω) is compared to n-1 angular thresholds 0<ai<a₂<...<a_n-i through equation (15) as follows: if Yj (ω) > βj(ω)/tan (ai) , then δ_j(ω)= εi = 1;

- if n is greater than 2 and β_j (ω) /tan (a_k-i) > Y_j (ω) ≥ β_j (ω) /tan (a_k) , then δ_j{ω)= ε_k, for k = 2, ... n-1; else, δ_j(ώ)= ε_n = 0.

The values ε_k are such that εi>ε₂> ... >ε_n>0, and these values may for instance equal: ε_k = cos (a_k-i) , for k = 2, ... n.

If n = 1, this comparison is limited to: if Yj (ω) > βj(ω)/tan (ai) , then δ_j(ω)=l; else, δ_j(ω)=0.

The subsequent steps S107 - Sill of the method may be identical to the steps explained above for the first and second embodiments of the invention.

According to one further embodiment of the invention, the sampling of the acoustic signals may be triggered only if the signals received by the analog to digital converter 8 are greater than a trigger threshold. Consequently, the device only samples relevant acoustic signals. This permits to reduce the interference sensibility. In this embodiment, the sampled signals ^₁(Z) and ^₂(Z) may be delayed compared to the real signals. As a matter of fact, to avoid that the trigger threshold prevents the record of the very beginning of acoustic signals, a delay device may be added, in order to record the waveform of the signal a few microseconds before the sampling step is triggered.

Claims

C L A I M S

1. A method for determining the position of impacts on an object, said object comprising: two acoustic sensors (SENSl, SENS2); - N predetermined active areas (Ia), whereby N is an integer at least equal to 1; said method comprising the steps of:

(a) receiving two acoustic signals ^₁(V) and ^₂(O originating respectively from said acoustic sensors (SENSl, SENS2) and generated by an impact received on said object;

(b) calculating a sample signature function:

SIG₈ ⁽«0=S»-S₂W> where

^{are the} respective Fourier transforms of ^₁(O and ^₂(O / and where * is the complex conjugate operator;

(c) comparing SIGv (⁶^) with N predetermined reference

signature functions SIGR- (^ω) each corresponding to a

predetermined active area j, for j from 1 to N; (d) determining an active area (Ia) in which the impact occurred, on the basis of the comparison of step (c) .

2. The method as claimed in claim 1, wherein each reference signature function equals:

where R₁X^ω) ^and R₂ (^ω) ^are Fourier transforms of acoustic signals /-,.(0 and γ>₂.(t) received by each of the respective acoustic sensors when an impact occurs on the predetermined area j .

3. The method as claimed in claim 1 or claim 2, wherein step (c) comprises the calculation of a similarity estimator a_} representing a function of the phase φ(COR₁(CO))Of CORj(ω) =SIG_R(<°)-SIG_S («0^* ■

4. The method as claimed in claim 3, wherein step (c) comprises the calculation of a function δ.(cό) for j from 1 to N, wherein δ_j(ώ)=ε_k, if φ(CORj(ω)) belongs to J_k , where ε_k is a predetermined value and J_k a corresponding angular interval for k from 1 to n, where n is an integer greater than 1.

5. The method as claimed in claim 4, wherein the values £", are not greater than 1.

6. The method as claimed in claim 4 or claim 5, wherein: if Iφ (COR_j(Co))I < al, then δ_j(ω)= ε_lr

if n is greater than 2 and a_k_i < | φ ( COR_j(ω))| ≤ a_k, then δ_j{ώ) ~ ε_k, for k=2... n-1 ,

if I φ ( CORJ(CU))I > a_n-i , then δ_j(ω) = ε_n, wherein a_k is increasing with k and ε_k is decreasing with k.

7. The method as claimed in claim 6, wherein: if Re(COR_j(CO)) ≥ 1Im(CORj(Co)) I / tan(ai), then δ_j{ω)= εi,

if n is greater than 2 and | Im(COR_j(co)) |/tan(a_k_i) > Re(CORj(ω))≥ | Im(COR_j(ω)) I /tan (a_k) , then δ_j(ω)= ε_kf for k=2... n-1, else , δ_j(ω) = ε_n, and wherein Re(CORj(ω)) is the real part of COR_j(ω) and Im(COR_j(ω)) is the imaginary part of COR_j(ω).

8. The method according to claim 6, wherein: - if Y_j (ω) ≥ βj(ω)/tan (ai) , then δ_j{ω)= E₁ = 1;

if n is greater than 2 and β_j (ω) /tan (a_k-i) > Y_j (ω) ≥ β_j (ω) /tan (a_k) , then δ_j{ω)= ε_k, for k = 2, ... n-1;

- else, δ_j(ω)= ε_n, and wherein: β_j(ω) = |l+{A_j(ω)/B_j(ω) } . { (D(ω) /C (ω) } | , Y_j (ω)=signBj (ω) .signC(ω) . [ {Aj (ω) /Bj (ω) }-{D(ω) /C(ω) }] , signB_j (ω) = 1 if Bj (ω) is positive and = -1 if B_j (ω) is negative, signC(ω)= 1 if C(ω) is positive and = -1 if C(ω) is negative,

A_j (ω) and B_j (co) are respectively the real part and the imaginary part of each reference signature function SIG_Rj (ω) ,

C(ω) and D(ω)are respectively the real part and the imaginary part of the complex conjugate sample signature function SIGs(ύ))*.

9. The method as claimed in any of claims 6 - 8, wherein εi = 1 and ε_n = 0.

10. The method as claimed in claim 9, wherein n is greater than 2 and ε_k = cos(a_k_i), for k = 2, ... n-1.

11. The method as claimed in any of claims 4 - 11, wherein a_j ,

where B is a frequency interval and K a constant.

12. The method as claimed in claim 11, wherein B = [ω_min,ω_max] and K is proportional to 1/ (ω_max . ω_min) .

13. The method as claimed in any of claims 3 - 12, wherein the active area j₀ in which the impact occurred is determined such that α_jα is the greatest similarity estimator among the N calculated similarity estimators α..

14. The method as claimed in claim 13, wherein it is concluded that the impact occurred in the area j_Q only if CC_j0 is considered as valid in step (d) .

15. The method of claim 14, wherein α_y. is considered as valid if α_jo is greater than a predetermined threshold of confidence (VAL) .

16. The method as claimed in claim 14 or claim 15, wherein it is concluded that there is no impact if α_Jo is not considered as valid in step (d) .

17. The method as claimed in any of the preceding claims, wherein step (a) begins if the acoustic signals

5,(0 and s₂(t) are above a predetermined trigger threshold.

18. The method as claimed in any of the preceding claims, wherein the acoustic signals ^₁(I) and ^₂(O in step

(a) are delayed relative to real audio signals.

19. The method as claimed in any of the preceding claims, wherein the Fourier transform is a Fast Fourier transform.

20. A device for determining the position of impacts on an object, comprising: two acoustic sensors adapted to be borne by said object for receiving acoustic signals ^₁(t) and ^₂(O generated by an impact on said object; memory means (10) comprising N reference signature functions, corresponding to N predetermined active areas (Ia) of said object, whereby N is an integer at least equal to 1; - calculation means (SlOl - S103) for calculating a sample signature function

where SS^ω) ^anci S₂W ^{are the} respective Fourier transforms of £,(0 and ^₂(O , and where * is the complex conjugate operator; comparison means (S104 - SIlO) for comparing SIGvW)

with N predetermined reference signature functions

SIG_RW ^for i ^{from 1 to N;} processing means (Sill) for determining an active area (Ia) in which the impact occurred, on the basis of results given by said the comparison means.

21. The device as claimed in claim 20, wherein each reference signature function equals:

where R₁ W and R_2jW are Fourier transforms of acoustic signals y_x-(i) and /*₂.(0 received by each of the respective acoustic sensors when an impact occurs on the predetermined area j .

22. The method as claimed in claim 1 or claim 2, wherein said comparison means (S104 - SIlO) are adapted to calculate a similarity estimator or_y. representing a function of the phase φ(COR_j(ω)) of

^•

23. The device as claimed in claim 22, wherein said comparison means (S104 - SIlO) are adapted to calculate a function δ_j(ώ) for j from 1 to N, wherein δ_j(ω)=ε_k, if φ(COR_j(ω)) belongs to J_k,

where £_k is a predetermined value and J_k a corresponding angular interval for k from 1 to n, where n is an integer greater than 1.

24. The device as claimed in claim 23, wherein the values £", are not greater than 1.

25. The device as claimed in claim 23 or claim 24, wherein said comparison means (S104 - SIlO) are adapted so that: if |φ(CORj(ω))| ≤ al, then δ_j(ω)= ε_lr

- if n is greater than 2 and a_k-i < |φ(COR_j(ω))| ≤ a_k, then δ_j{ω)= ε_k, for k=2... n-1,

if I φ (COR)(ω))| > a_n-i, then δ_j(ω)= ε_n, wherein a_k is increasing with k and ε_k is decreasing with k.

26. The device as claimed in claim 25, wherein said comparison means (S104 - SIlO) are adapted so that: if Re(CORj(Co)) ≥ |Im(COR_j(ω)) | / tan(ai), then δ_j(ω)= ει,

if n is greater than 2 and |Im(COR_j(ω)) |/tan(a_k-i) > Re (COR_j(co)) ≥ Hm(COR_j(Co)) I /tan (a_k) , then δ_j(ω)= ε_k, for k=2... n-1,

- else, δ_j(ω)= E_n, and wherein Re (COR_j(ω)) is the real part of COR_j(ω) and Im(COR_j(ω)) is the imaginary part of COR_j(ω).

27. The device according to claim 25, wherein said comparison means (S104 - SlIO) are adapted so that: - if Yj (ω) > βj (ω) /tan (ai) , then δ_j(ω)= εi = 1;

if n is greater than 2 and βj (ω) /tan (a_k-i) > Y_j (ω) ≥ βj (ω) /tan(a_k) , then δ_j{ω)= ε_k/ for k = 2, ... n-1;

- else, δ_j(ω)= ε_n, and wherein: β_j(ω)= |l+{A_j(ω)/B_j(ω) } . { (D(ω) /C (ω) } | , Y_j (ω)=signBj (ω) .signC(ω) . [ {A_j (ω) /B_j (ω) }-{D(ω)/C(ω) }], signBj (ω) = 1 if Bj (ω) is positive and = -1 if B_j (ω) is negative, signC(ω)= 1 if C(co) is positive and = -1 if C(ω) is negative,

Aj (ω) and Bj (ω) are respectively the real part and the imaginary part of each reference signature function SIG_Rj(ω),

C(ω) and D(ω)are respectively the real part and the imaginary part of the sample signature function and SIGs(ω) .

28. The device as claimed in any of claims 25 - 27, wherein εi = 1 and ε_n = 0.

29. The device as claimed in claim 28, wherein n is greater than 2 and ε* = cos(a_k-i), for k = 2, ... n-1.

30. The device as claimed in any of claims 23 - 30, wherein said comparison means (S104 - SIlO) are adapted so that: a_j =K.[δ_j(ω).dω , where B is a frequency interval and K a constant.

31. The device as claimed in any of claims 22 - 30, wherein the processing means (Sill) are adapted to determine the active area j₀ in which the impact occurred such that a_} is the greatest similarity estimator among the N calculated similarity estimators α_y..

32. The device as claimed in claim 31, wherein the processing means (Sill) are adapted to determine that the impact occurred in the area J₀ only if a_Jo is considered as valid.

33. The device of claim 32, wherein the processing means (Sill) are adapted to determine that a_Jo is valid if a. is greater than a predetermined threshold of confidence

(VAL) .

34. The device as claimed in claim 32 or claim 33, wherein the processing means (Sill) are adapted to determine that there is no impact if a_Jo is not considered as valid.