CN105223567A - A kind of robust wideband Adaptive beamformer method being applied to ultrasonic imaging - Google Patents

Abstract

The present invention relates to a kind of robust wideband Adaptive beamformer method being applied to ultrasonic imaging, comprising: obtained ultrasonic echo data is converted to frequency domain by time domain, then carry out the division of frequency band; Calculate the covariance matrix of each sub-band after described frequency band division; According to the described covariance matrix obtained, Feature Space Decomposing method is utilized to be projected in signal subspace by the steering vector of supposition; Real steering vector is estimated by convex optimization method; Utilize the ultimate principle of Minimum Variance method, obtain the Wave beam forming result in each sub-band space; The Wave beam forming result in described each sub-band space is superposed, and is transformed into time domain, obtain the Output rusults for final imaging.The present invention makes full use of the broadband property feature of ultrasound echo signal, is reconstructed in frequency domain to the steering vector under true environment, can possess comparatively sane performance under the condition of phase distortion, and improves the ultrasonic imaging quality under nonhomogeneous media simultaneously.

Description

A kind of robust wideband Adaptive beamformer method being applied to ultrasonic imaging

Technical field

The present invention relates to the field such as medical ultrasound diagnosis and Ultrasonic NDT, refer more particularly to a kind of robust wideband Adaptive beamformer method being applied to ultrasonic imaging.

Background technology

Generally employ the Beamforming Method of time delay superposition in traditional medical ultrasound diagnosis and ultrasonic nondestructive testing, this method is simple and reliable, is easy to realize, but performance is more general.For improving the quality of ultrasonic imaging, minimum variance Adaptive beamformer method receives to be paid close attention to widely, but this method faces two following subject matters: the minimum variance Beamforming Method of (1) classics proposes according to the model of narrow band signal, and the echoed signal in ultrasonic imaging field generally belongs to the category of broadband signal, so directly minimum variance Beamforming Method to be applied in ultrasonic imaging and to fail fully to show the advantage of former method; (2) often there is phase distortion in the imaging process of nonhomogeneous media, the estimation of the covariance matrix in minimum variance Beamforming Method and steering vector is caused to there is certain error, then have impact on the robustness of the method, be difficult to the performance of realizing ideal in comparatively complicated applied environment.

The robust adaptive beamforming method of current existence, mostly be centered around in the reconstruct of covariance matrix and the estimation of steering vector, as in minimum variance Beamforming Method, normal use diagonal angle loading technique improves the robustness of covariance matrix, this method is simple, choosing of diagonal angle this parameter of load factor, has important impact to performance.Adopting front and back to the space smoothing mode combined to ultrasound echo signal decoherence, without the need to using diagonal angle to load the robustness that can ensure covariance matrix to a certain extent, also can improve the contrast of final imaging simultaneously.But said method all concentrates on process in time domain, does not fully take into account the broadband property feature of ultrasound echo signal, also fails effectively to estimate the steering vector of mismatch under phase distortion condition, cause final performance boost comparatively limited.

Summary of the invention

The present invention is directed to the deficiencies in the prior art, minimum variance Adaptive beamformer method is improved, make full use of the broadband property feature of ultrasound echo signal, in frequency domain, the steering vector under true environment is reconstructed, comparatively sane performance can be possessed under the condition of phase distortion, and improve the ultrasonic imaging quality under nonhomogeneous media simultaneously.

For achieving the above object, the invention provides a kind of robust wideband Adaptive beamformer method being applied to ultrasonic imaging, be applied in the ultrasound transducer array of an equally spaced M array element, the method is mainly used in the stage of the echo data received being carried out to Wave beam forming process, comprises the following steps:

Obtained ultrasonic echo data is converted to frequency domain by time domain, then carries out the division of frequency band;

Calculate the covariance matrix of each sub-band after described frequency band division;

According to the described covariance matrix obtained, Feature Space Decomposing method is utilized to be projected in signal subspace by the steering vector of supposition;

Real steering vector is estimated by convex optimization method;

Utilize the ultimate principle of Minimum Variance method, obtain the Wave beam forming result in each sub-band space;

The Wave beam forming result in described each sub-band space is superposed, and is transformed into time domain, obtain the Output rusults for final imaging.

Further, described obtained ultrasonic echo data is converted to frequency domain by time domain, then carries out the partiting step of frequency band, comprising:

Short time discrete Fourier transform is carried out to echo data received by each array element corresponding of each root sweep trace under B-mode ultrasonic imaging, the frequency domain data obtained after conversion is divided into K sub-band.

Further, described obtained ultrasonic echo data is converted to frequency domain by time domain, the partiting step then carrying out frequency band comprises:

Design the width of the window function of described short time discrete Fourier transform according to the pulse width of launching in advance, then carry out the division in echo data space;

" zero padding " process is carried out to each segment data after described echo data spatial division, is then converted to frequency domain by time domain;

Described frequency domain data is carried out to the division of frequency band.

Further, the covariance matrix step of each sub-band after the described frequency band division of described calculating, comprising:

Carry out space smoothing process to the data of each sub-band, be wherein one group with L array element and calculate covariance matrix, be divided into M-L+1 group, the formula of calculating is as follows:

$\hat{R}\left({\mathrm{\ω}}_i\right)=\frac{1}{M-L+1}\underset{l=1}{\overset{M-L+1}{\Σ}}{x}_d^l\left({\mathrm{\ω}}_i\right)x_d^l{\left({\mathrm{\ω}}_i\right)}^H$

Wherein, covariance matrix, the echo data on frequency domain, be conjugate transpose, i=1,2 ..., K, K are the number of sub-band.

Further, the steering vector of supposition is projected in step in signal subspace by the described Feature Space Decomposing method that utilizes, and comprising:

By following formula, the covariance matrix obtained is decomposed at feature space:

$\hat{R}={\mathrm{U\ΛU}}^H=U_s{\mathrm{\Λ}}_s{U_s}^H+U_p{\mathrm{\Λ}}_p{U_p}^H$

Wherein Λ=diag [λ ₁, λ ₂..., λ _num..., λ _l], and eigenwert (λ ₁>=λ ₂>=... λ _num>=...>=λ _l), U=[u ₁, u ₂..., u _num... u _l] be the eigenvector corresponded, Selecting All Parameters β, between 0.1 ~ 0.5, meets λ _num>=β λ ₁;

Whole feature space is divided into signal subspace Us and noise subspace Up two parts.

Utilize the signal subspace Us obtained, the steering vector of calculation assumption projection in signal subspace:

$a=U_s{U_s}^H\stackrel{\‾}{a}$

Wherein, Us is signal subspace, U _s ^hfor the conjugate transpose of Us, assuming that steering vector in element be all taken as " 1 ".

Further, describedly estimate real steering vector by convex optimization method, comprising:

Steering vector a under utilizing convex optimization method to estimate truth, computing formula is as follows::

$\underset{a}{\min }{a}^H{\hat{R}}^{-1}a$

$\begin{array}{cc}s.t.& \left|\right|a-\tilde{a}\left|\right|\≤\mathrm{\ϵ}\end{array}$

Wherein, the inverse of covariance matrix, steering vector be the supposition projection of steering vector in signal subspace, parameter ε is set error range, generally between between.

Further, the described ultimate principle utilizing Minimum Variance method, obtains the Wave beam forming result in each sub-band space, comprising:

According to the steering vector under the covariance matrix obtained and truth, the first principles computations of minimum variance Wave beam forming is utilized to go out one group of optimal weight vector;

Particularly, calculate by following formula:

$w_{MV}\left({\mathrm{\ω}}_i\right)=\frac{{\hat{R}}^{-1}\left({\mathrm{\ω}}_i\right)a}{a^H{\hat{R}}^{-1}\left({\mathrm{\ω}}_i\right)a}$

Wherein w _mV(ω _i) be optimal vector, be the inverse of the covariance matrix obtained, steering vector a is the result under the truth estimated;

The Output rusults of the Wave beam forming of each subband data is gone out according to the optimal weight vector obtained and following formulae discovery:

$y\left({\mathrm{\ω}}_i\right)=w_{MV}^H\left({\mathrm{\ω}}_i\right)\·x_d\left({\mathrm{\ω}}_i\right)$

Wherein, w _mV(ω _i) conjugate transpose, y (ω _i) be the Output rusults of the Wave beam forming of i-th sub-band, x _d(ω _i) be echo data on frequency domain.

Further, the described Wave beam forming result to described each subband data superposes, and obtains the Wave beam forming Output rusults step of domain space, comprising:

Add up to the result of the Wave beam forming in each sub-band space in every one piece of data, obtain the Wave beam forming Output rusults of domain space, the formula of calculating is as follows:

$Y\left(\mathrm{\ω}\right)=\underset{i=1}{\overset{K}{\Σ}}y\left({\mathrm{\ω}}_i\right)$

Wherein, Y (ω) is the Wave beam forming Output rusults of domain space, y (ω _i) be the Output rusults of the Wave beam forming of i-th sub-band.

The invention has the advantages that:

The echo data of time domain to be transformed in each sub-band by short time discrete Fourier transform and to process by the present invention, takes full advantage of the broadband feature of ultrasound echo signal, can obtain more preferably performance; In each frequency band, the steering vector of supposition is projected in signal subspace simultaneously, and adopts convex optimization method to be reconstructed, improve the robustness of algorithm under phase distortion condition.

Accompanying drawing explanation

The schematic flow sheet that Fig. 1 provides for the embodiment of the present invention;

What Fig. 2 provided for the embodiment of the present invention carries out the pretreated schematic diagram of time delay to the echo data on original time domain;

The theory diagram obtaining final Output rusults that Fig. 3 provides for the embodiment of the present invention;

Fig. 4 is for each sub-band space that the embodiment of the present invention provides carrying out the processing procedure of Wave beam forming.

Embodiment

Below in conjunction with drawings and Examples, technical scheme of the present invention is described in further detail.

A kind of Adaptive beamformer method flow schematic diagram that Fig. 1 provides for the embodiment of the present invention, the embodiment of the present invention is applied to the ultrasonic imaging under B-mode.Under the ultrasound transducer array of an equally spaced M array element is operated in B-mode imaging linear scanning mode, Wave beam forming process is carried out to received echo data, each root sweep trace correspond to the Received signal strength (namely M is array element number) of M array element, and the number of samples of each array element is N, the echo data that all array elements that each root sweep trace is corresponding under B-mode ultrasonic imaging receive can describe with the matrix of M*N.It should be noted that, the echo data in these time domains is through the pre-service of time delay, and Fig. 2 gives time delay pretreated schematic diagram, namely with the x that pre-service obtains in following steps _chandling object by default, it is to be noted that echo data can not ensure to obtain perfectly " alignment ", and the present invention is also without the need to making extra hypothesis to the time delay pre-processed results of echo data due to the impact of nonhomogeneous media simultaneously.Fig. 3 is the theory diagram obtaining final Output rusults that the embodiment of the present invention provides, shown in Fig. 1 and Fig. 3, the known embodiment of the present invention Wave beam forming disposal route comprise step 101-103:

Step 101, is converted to frequency domain by obtained ultrasonic echo data by time domain, then carries out the division of frequency band;

Concrete, the echo data that all array elements under B-mode ultrasonic imaging in each root sweep trace receive can describe with the matrix of M*N, wherein M is array element number, N is number of samples, it should be noted that, this echo data has completed time delay pre-service, carries out short time discrete Fourier transform to this echo data, and the frequency domain data obtained after conversion is divided into K sub-band, for the Wave beam forming in each sub spaces completes necessary preparation.

Be described in further detail the disposal route of described step 101 below:

Step 1011, designs the width of the window function of described short time discrete Fourier transform in advance, then carries out the division in echo data space according to the pulse width of launching;

Particularly, design the width Nl of suitable window function according to the pulse width of launching, whole data space is divided into Ng section; Wherein, N is number of samples, and Ng=N/Nl also gets its integral part, and namely each section of corresponding data space size is M × Nl.Need to further illustrate, for ensureing the longitudinal frame of final imaging, N1 generally can not exceed exomonental width.

Step 1012, carries out " zero padding " process to each segment data after described echo data spatial division, is then converted to frequency domain by time domain;

Particularly, (Nf-Nl) individual " 0 " is added after echo data in the time domain corresponding to each array element, then carry out Fourier transform process to every one piece of data, be transformed into frequency domain by the echo data in time domain, the echo data after Short Time Fourier Transform will have Ng section;

In processing procedure, in order to the convenient calculating of short time discrete Fourier transform, Nf generally gets the exponential depth of 2, and meets Nf>2Nl, thus the size in every one piece of data space becomes M*Nf, each section of frequency domain data namely obtained after conversion is all the matrix of M*Nf;

Step 1013, carries out the division of frequency band to described frequency domain data;

Particularly, above-mentioned steps 1012 obtains the frequency domain data of Ng section M*Nf altogether, to the frequency domain data obtained, carries out frequency band division, is K sub-band by each section of frequency band division, wherein, and Nl<K<Nf/2.

Step 102, carries out Wave beam forming process to each sub-band space after described frequency band division, obtains the Wave beam forming result in described each sub-band space;

Particularly, as shown in Figure 4, in each divided sub-band, after the covariance matrix obtaining each sub-band, first by the method for Feature Space Decomposing, the steering vector of supposition is projected in signal subspace, then estimate the steering vector under practical situation by convex optimization method, then utilize the ultimate principle of Minimum Variance method, obtain forming Output rusults to each sub-band spatial beams.

Below with i-th (i=1,2 ..., K) and individual sub-band is example, introduces the processing procedure of step 102 in detail:

Step 1021, calculates the covariance matrix of each sub-band;

Particularly, the size of data corresponding to i-th sub-band is M × 1, first carries out space smoothing process to the data that dimension is M × 1, is wherein one group with L array element and calculates covariance matrix, and be divided into M-L+1 group, the formula of calculating is as follows:

$\hat{R}\left({\mathrm{\&omega;}}_i\right)=\frac{1}{M-L+1}\underset{l=1}{\overset{M-L+1}{\&Sigma;}}{x}_d^l\left({\mathrm{\&omega;}}_i\right)x_d^l{\left({\mathrm{\&omega;}}_i\right)}^H$

Wherein, covariance matrix, the echo data on frequency domain, be conjugate transpose, i=1,2 ..., K, K are the number of sub-band.

Step 1022, decomposes at feature space the covariance matrix obtained in step 1021;

Particularly, decompose according to following formula:

$\hat{R}={\mathrm{U\&Lambda;U}}^H=U_s{\mathrm{\&Lambda;}}_s{U_s}^H+U_p{\mathrm{\&Lambda;}}_p{U_p}^H$

Wherein Λ=diag [λ ₁, λ ₂..., λ _num..., λ _l], and eigenwert is according to the arrangement of descending order, i.e. (λ ₁>=λ ₂>=... λ _num>=...>=λ _l), and U=[u ₁, u ₂..., u _num... u _l] be then the eigenvector corresponded, choose suitable parameter beta, meet λ _num>=β λ ₁, then whole feature space is divided into signal subspace Us and noise subspace Up two parts.For ensureing the sane performance of algorithm, the parameter beta of selection is between 0.1 ~ 0.5;

Step 1023, the component of signal feature space utilizing step 1022 to obtain, the steering vector of calculation assumption projection in signal subspace;

Particularly, calculate by following formula:

$a=U_s{U_s}^H\stackrel{\&OverBar;}{a}$

Wherein, Us is signal subspace, U _s ^hfor the conjugate transpose of Us, assuming that steering vector in element be all taken as " 1 ";

Step 1024, the steering vector a under utilizing convex optimization method to estimate truth;

Particularly, calculate by following formula:

$\underset{a}{\min }{a}^H{\hat{R}}^{-1}a$

$\begin{array}{cc}s.t.& \left|\right|a-\tilde{a}\left|\right|\&le;\mathrm{\&epsiv;}\end{array}$

Wherein inverse by the covariance matrix obtained in step 1021, steering vector be the projection of supposition steering vector in signal subspace obtained in step 1023, parameter ε is set error range, its value generally between between.

Step 1025, according to the steering vector of the covariance matrix obtained and reconstruct, utilizes the first principles computations of minimum variance Wave beam forming to go out one group of optimal weight vector;

Particularly, calculate by following formula:

$w_{MV}\left({\mathrm{\&omega;}}_i\right)=\frac{{\hat{R}}^{-1}\left({\mathrm{\&omega;}}_i\right)a}{a^H{\hat{R}}^{-1}\left({\mathrm{\&omega;}}_i\right)a}$

Wherein w _mV(ω _i) be optimal weight vector, be the inverse of the covariance matrix obtained in step 1021, steering vector a is the estimated result obtained in step 1024.

Step 1026, goes out the Output rusults of the Wave beam forming in i-th sub-band space according to the optimal weight vector obtained and following formulae discovery:

$y\left({\mathrm{\&omega;}}_i\right)=w_{MV}^H\left({\mathrm{\&omega;}}_i\right)\&CenterDot;x_d\left({\mathrm{\&omega;}}_i\right)$

Wherein, w _mV(ω _i) conjugate transpose, y (ω _i) be the Output rusults of the Wave beam forming of i-th sub-band, x _d(ω _i) be echo data on frequency domain.

Step 1027, according to above-mentioned method to calculate successively under B-mode imaging each sub-band in each root sweep trace (i=1,2,3 ..., K) Wave beam forming Output rusults;

Step 103, adds up the Wave beam forming Output rusults that each sub-band space obtains, and is transformed into time domain, obtain the Output rusults for final imaging.

Particularly, the disposal route of described step 103 comprises:

Step 1031, adds up the Output rusults of the Wave beam forming of the sub-band of the K in every one piece of data, and obtain the Wave beam forming Output rusults of frequency domain, computing formula is as follows:

$Y\left(\mathrm{\&omega;}\right)=\underset{i=1}{\overset{K}{\&Sigma;}}y\left({\mathrm{\&omega;}}_i\right)$

Wherein, Y (ω) is the Wave beam forming Output rusults of domain space, y (ω _i) be the Output rusults of the Wave beam forming of i-th sub-band.

Obtain the Wave beam forming Output rusults of domain space, also need to carry out follow-up process, be shown as picture for final:

Step 1032, is transformed into time domain by the Wave beam forming Output rusults of domain space;

Particularly, the Wave beam forming Output rusults of the domain space in every one piece of data is carried out inverse fourier transform, and the Wave beam forming Output rusults by domain space is converted to time domain, obtains final Output rusults, in processing procedure, the size of every segment data is taken as M × Nl.

Step 1033, repeats above-mentioned process, calculates the Output rusults of all Ng sections, and the size of data that the Wave beam forming making each root sweep trace corresponding exports is 1 × N, is shown as picture for final.

The present invention relates to a kind of robust wideband Adaptive beamformer method being applied to ultrasonic imaging, the main points of the method comprise: the basic thought utilizing minimum variance Beamforming Method, improve the performance of Beam-former in conjunction with short time discrete Fourier transform, Feature Space Decomposing and steering vector method of estimation.First obtained echoed signal is converted to frequency domain by time domain, and be divided into some subbands according to the band characteristics of ultrasonic signal, then on each subband, Wave beam forming process is done respectively, namely by the method for Feature Space Decomposing, the steering vector of supposition is projected in signal subspace, then the steering vector under practical situation is estimated by convex optimization method, then the ultimate principle of Minimum Variance method is utilized, obtain the result to each sub-band space, and carry out superposing the Wave beam forming Output rusults obtaining whole frequency domain, the data finally utilizing inverse fourier transform to transfer in time domain are shown as picture for last.The present invention can improve the robustness of Beamforming Method under nonhomogeneous media environment, and significantly improves the quality of ultrasonic imaging simultaneously, can be applied in the actual demand such as medical ultrasound diagnosis and Ultrasonic NDT.

Professional should recognize further, in conjunction with unit and the algorithm steps of each example of embodiment disclosed herein description, can realize with electronic hardware, computer software or the combination of the two, in order to the interchangeability of hardware and software is clearly described, generally describe composition and the step of each example in the above description according to function.These functions perform with hardware or software mode actually, depend on application-specific and the design constraint of technical scheme.Professional and technical personnel can use distinct methods to realize described function to each specifically should being used for, but this realization should not thought and exceeds scope of the present invention.

The software module that the method described in conjunction with embodiment disclosed herein or the step of algorithm can use hardware, processor to perform, or the combination of the two is implemented.Software module can be placed in the storage medium of other form any known in random access memory (RAM), internal memory, ROM (read-only memory) (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technical field.

Above-described embodiment; object of the present invention, technical scheme and beneficial effect are further described; be understood that; the foregoing is only the specific embodiment of the present invention; the protection domain be not intended to limit the present invention; within the spirit and principles in the present invention all, any amendment made, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (9)

1. be applied to a robust wideband Adaptive beamformer method for ultrasonic imaging, it is characterized in that, the method comprises:

Obtained ultrasonic echo data is converted to frequency domain by time domain, then carries out the division of frequency band;

Calculate the covariance matrix of each sub-band after described frequency band division;

According to the described covariance matrix obtained, Feature Space Decomposing method is utilized to be projected in signal subspace by the steering vector of supposition;

Real steering vector is estimated by convex optimization method;

Utilize the ultimate principle of Minimum Variance method, obtain the Wave beam forming result in each sub-band space;

The Wave beam forming result in described each sub-band space is superposed, and is transformed into time domain, obtain the Output rusults for final imaging.

2. method according to claim 1, is characterized in that, described obtained ultrasonic echo data is converted to frequency domain by time domain, then carries out the partiting step of frequency band, comprising:

Short time discrete Fourier transform is carried out to echo data received by each array element corresponding of each root sweep trace under B-mode ultrasonic imaging, the frequency domain data obtained after conversion is divided into K sub-band.

3. method according to claim 2, is characterized in that, described obtained ultrasonic echo data is converted to frequency domain by time domain, and the partiting step then carrying out frequency band comprises:

Design the width of the window function of described short time discrete Fourier transform according to the pulse width of launching in advance, then carry out the division in echo data space;

" zero padding " process is carried out to each segment data after described echo data spatial division, is then converted to frequency domain by time domain;

Described frequency domain data is carried out to the division of frequency band.

4. according to the method in claims 1 to 3 described in any one, it is characterized in that, it is pretreated that described ultrasonic echo data is through time delay.

5. method according to claim 1, is characterized in that, the covariance matrix step of each sub-band after the described frequency band division of described calculating, comprising:

Carry out space smoothing process to the data of each sub-band, be wherein one group with L array element and calculate covariance matrix, be divided into M-L+1 group, the formula of calculating is as follows:

$\hat{R}\left({\mathrm{\&omega;}}_i\right)=\frac{1}{M-L+1}\underset{l=1}{\overset{M-L+1}{\&Sigma;}}{x}_d^l\left({\mathrm{\&omega;}}_i\right)x_d^l{\left({\mathrm{\&omega;}}_i\right)}^H$

Wherein, covariance matrix, the echo data on frequency domain, be conjugate transpose, i=1,2 ..., K, K are the number of sub-band.

6. method according to claim 1, is characterized in that, the steering vector of supposition is projected in step in signal subspace by the described Feature Space Decomposing method that utilizes, and comprising:

By following formula, the covariance matrix obtained is decomposed at feature space:

$\hat{R}={\mathrm{U\&Lambda;U}}^H=U_s{\mathrm{\&Lambda;}}_s{U_s}^H+U_p{\mathrm{\&Lambda;}}_p{U_p}^H$

Wherein Λ=diag [λ ₁, λ ₂..., λ _num..., λ _l], and eigenwert (λ ₁>=λ ₂>=... λ _num>=...>=λ _l), U=[u ₁, u ₂..., u _num... u _l] be the eigenvector corresponded, Selecting All Parameters β, between 0.1 ~ 0.5, meets λ _num>=β λ ₁;

Whole feature space is divided into signal subspace Us and noise subspace Up two parts.

Utilize the signal subspace Us obtained, the steering vector of calculation assumption projection in signal subspace:

$a=U_s{U_s}^H\stackrel{\&OverBar;}{a}$

Wherein, Us is signal subspace, U _s ^hfor the conjugate transpose of Us, assuming that steering vector in element be all taken as " 1 ".

7. method according to claim 1, is characterized in that, describedly estimates real steering vector by convex optimization method, comprising:

Steering vector a under utilizing convex optimization method to estimate truth, computing formula is as follows: _:

$\underset{a}{\min }{a}^H{\hat{R}}^{-1}a$

$\begin{array}{cc}s.t.& ||a-\tilde{a}||\&le;\mathrm{\&epsiv;}\end{array}$

Wherein, the inverse of covariance matrix, steering vector be the supposition projection of steering vector in signal subspace, parameter ε is set error range, generally between between.

8. method according to claim 1, is characterized in that, the described ultimate principle utilizing Minimum Variance method, obtains the Wave beam forming result in each sub-band space, comprising:

According to the steering vector under the covariance matrix obtained and truth, the first principles computations of minimum variance Wave beam forming is utilized to go out one group of optimal weight vector;

Particularly, calculate by following formula:

$w_{MV}\left({\mathrm{\&omega;}}_i\right)=\frac{{\hat{R}}^{-1}\left({\mathrm{\&omega;}}_i\right)a}{a^H{\hat{R}}^{-1}\left({\mathrm{\&omega;}}_i\right)a}$

Wherein w _mV(ω _i) be optimal weight vector, be the inverse of the covariance matrix obtained, steering vector a is the result under the truth estimated;

The Output rusults of the Wave beam forming of each subband data is gone out according to the optimal weight vector obtained and following formulae discovery:

$y\left({\mathrm{\&omega;}}_i\right)=w_{MV}^H\left({\mathrm{\&omega;}}_i\right)\&CenterDot;x_d\left({\mathrm{\&omega;}}_i\right)$

Wherein, w _mV(ω _i) conjugate transpose, y (ω _i) be the Output rusults of the Wave beam forming of i-th sub-band, x _d(ω _i) be echo data on frequency domain.

9. according to the method described in claim 1, it is characterized in that, the described Wave beam forming result to described each subband data superposes, and obtains the Wave beam forming Output rusults step of domain space, comprising:

Add up to the result of the Wave beam forming in each sub-band space in every one piece of data, obtain the Wave beam forming Output rusults of domain space, the formula of calculating is as follows:

$Y\left(\mathrm{\&omega;}\right)=\underset{i=1}{\overset{K}{\&Sigma;}}y\left({\mathrm{\&omega;}}_i\right)$

Wherein, Y (ω) is the Wave beam forming Output rusults of domain space, y (ω _i) be the Output rusults of the Wave beam forming of i-th sub-band.