CN105182328A - Ground penetrating radar underground target positioning method - Google Patents

Abstract

The invention relates to a ground penetrating radar underground target positioning method. The method comprises the steps that 1) two-dimensional empirical mode decomposition is carried out on detection echo data of a ground penetrating radar to acquire K two-dimensional empirical mode function components IMF with orderly descending frequency and a residual; 2) the mean of the first M (M</=K) two-dimensional empirical mode function components is used as the characteristic value of the detection echo data; 3) the extreme point of the characteristic value of the detection echo data is acquired and is used as the estimation value of the vertex position of an underground target; 4) the underground propagation velocity of an electromagnetic wave is estimated; and 5) according to the estimation value of the vertex position of the underground target and the underground propagation velocity of the electromagnetic wave, a ground penetrating radar hyperbolic mathematical model is used to carry out hyperbolic fitting to position the underground target. According to the method provided by the invention, the target information is completely kept; a clutter suppression effect is improved; and the accuracy of target positioning is improved.

Description

A kind of ground penetrating radar buried target localization method

Technical field

The present invention relates to ground penetrating radar detection field, be specifically related to a kind of ground penetrating radar buried target localization method.

Background technology

Ground penetrating radar is the one effective shallow underground target detection technology developed rapidly nearly decades, it is a kind of non-destructive detection means, there is the plurality of advantages such as speed of detection is fast, resolution is high, flexible to operation, detection cost is low, be widely used in buried target, as detection and the location of cavity, pipeline, land mine etc.

The two-dimentional echo data of ground penetrating radar detection is called B-Scan data, and it is the data basis of follow-up Radar Signal Processing, target identification and decipher, and ground penetrating radar Technology for Target Location also will based on B-Scan data.To the accurate positioning effects of realize target maximum be " clutter " in ground penetrating radar B-Scan data.Ground penetrating radar clutter can regard the various echoes except target echo as, generally includes antenna direct wave, echo and the echo that produces of pseudo-target etc. that earth's surface echo, underground non-uniform medium produce.Ground penetrating radar clutter makes to become difficulty to the accurate detection of buried target, especially for shallow-layer Baseband Pules, target echo is more weak composition compared with the echo of earth's surface, and time delay between target echo and earth's surface echo is very little, target echo easily flood by this kind of clutter of the strong echo in earth's surface.Therefore ground penetrating radar clutter reduction realizes the top priority that ground penetrating radar target accurately locates.

Common localization method mainly extracts based on the hyperbolic curve of B scan image, carries out speed calculate target depth according to the hyperbolic curve extracted.Mainly contain: based on neural network to hyp extraction, need more data to train, not easily realize on-line checkingi; The mode identification method of Bian fuzzy clustering, for the shallow-layer detection that metal pipe line and nonmetal pipe line all may exist, easily produces false-alarm, and easily misses nonmetallic pipe line target.Based on the method for Iamge Segmentation and Hough transformation, when being applied in shallow-layer detection pipeline, can not effectively distinguish stronger clutter and target echo; Time method based on Iamge Segmentation and template matches is applied in shallow-layer detection pipeline, because the size of caliber may be changeable, thus the masterplate of correspondence is also more, causes algorithm longer for operation time; Based on morphologic curve detection, be carry out detection according to the gray-scale value of image to judge, but can judge order target area obtain being many curves, carrying out next step calculating also needs to process curve.

Summary of the invention

The invention provides a kind of ground penetrating radar buried target localization method, be intended to solve the complicated and problem that positioning precision is not high of object localization method of the prior art.

For solving the problems of the technologies described above, technical scheme of the present invention is:

1) echo data is detected to the B-Scan of ground penetrating radar and carry out two-dimensional empirical mode decomposition, obtain two-dimensional empirical modal function component IMF and 1 residual error that K frequency is successively decreased successively;

2) using the eigenwert of the average of front M (M≤K) two-dimensional empirical modal function component as detection echo data;

3) extreme point of the eigenwert of described detection echo data is obtained, as the estimated value of buried target vertex position;

4) velocity of propagation of electromagnetic wave in underground is estimated;

5) according to estimated value and the velocity of propagation of electromagnetic wave in underground of described buried target vertex position, utilize ground penetrating radar hyperbolic curve mathematical model, carry out hyperbolic fit, complete the location of buried target position.

Described step 1) in the detailed process that the detection echo data of ground penetrating radar carries out two-dimensional empirical mode decomposition be:

A) the detection echo data I of ground penetrating radar is first determined _resall extreme points, concrete employing eight neighborhood method determines I _resthe maximum value that image is all and minimal value;

B) to the detection echo data I of ground penetrating radar _resall extreme points utilize radial basis function to carry out interpolation, the maximum point after interpolation and minimum point use E respectively _iand E _srepresent, after carrying out curve fitting, obtain detection echo data I _resupper and lower envelope;

The concrete form of radial basis function RBF is:

$s\left(x\right)=p_m\left(x\right)+\underset{i=1}{\overset{N}{\&Sigma;}}{\mathrm{\&lambda;}}_i\mathrm{\&Phi;}\left(\right||x-x_i|\left|\right)---\left(1\right)$

Wherein: s is radial basis function (RBF), p _mpolynomial of lower degree, as linear or secondary or d variable m ^thpolynomial expression, || || represent euclideam norm.λ _ibe RBF coefficient, Φ is real-valued function, the center of the radial basis function RBF that is often known as.

C) average of upper and lower envelope is asked

E _M＝(E _I+E _S)/2；(2)

D) from original detection echo data I _resin deduct E _m, obtain new detection echo data

$I_{res}^{11}=I_{res}-E_M;---\left(3\right)$

E) judge according to IMF decision condition whether be an IMF, if an IMF, make first two-dimensional empirical modal function component (IMF) for residual error otherwise, use replace I _res, repeat step a) ~ d) until judge be an IMF, make first two-dimensional empirical modal function component (IMF) for residual error repetition like this, until two-dimensional empirical modal function component IMF and 1 residual error obtaining that K frequency successively decrease successively.

Described IMF decision condition is setting SD threshold value,

$SD=\underset{m=0}{\overset{M}{\&Sigma;}}\underset{n=0}{\overset{N}{\&Sigma;}}\&lsqb;\frac{{|I_{{\mathrm{res}}_{i(j-1)}}(m,n)-I_{{\mathrm{res}}_{ij}}(m,n)|}^2}{I_{{\mathrm{res}}_{i(j-1)}^2}(m,n)}\&rsqb;---\left(4\right)$

Wherein, with for passing through i-th ^ththe double attenuation results of individual pattern, represent i-th ^ththe numerical value of the capable n row of m of the jth time decay of individual Mode Decomposition, M, N represent line number and the columns of two-dimentional ground penetrating radar image.Preset a threshold value T in reality, stop iteration when SD is less than this threshold value, namely judge an IMF.

Described step 3) in learn that target echo has Hyperbolic Feature according to ground penetrating radar principle, the ordinate on hyperbolic curve summit represents the shortest echo time delay, namely nearest at this measuring point ground penetrating radar distance objective.Therefore, scan by column the eigenwert of the detection echo data chosen, choose the minimum value of ordinate, determine the ordinate on hyperbolic curve summit.Hyp horizontal ordinate just represents horizontal level corresponding to target.Described step 4) in adopt frequency wave beam deflection method and in conjunction with the minimum entropy technology estimation velocity of propagation of electromagnetic wave in underground.

Described step 5) in ground penetrating radar hyperbolic curve mathematical model be:

$\frac{t^2}{t_0^2}-\frac{4{(x-x_0)}^2}{v^2{t}_0^2}=1---\left(5\right)$

Wherein, x represents aerial position, x ₀represent the horizontal coordinate of representative points position, v represents the velocity of propagation of electromagnetic wave in underground, t ₀expression aerial position is x ₀target reflection echo time delay, t represents that aerial position is the target reflection echo time delay of x.

First ground penetrating radar buried target localization method of the present invention carries out two-dimensional empirical mode decomposition to the detection echo data of ground penetrating radar, obtain several single-component signals, then detection echo data eigenwert is extracted according to single-component signal, estimation representative points position, then the velocity of wave and ground penetrating radar principle that estimate is combined, carry out hyperbolic fit, complete target localization.The method promotes clutter recognition effect while more complete reservation target information, improves the precision of target localization.

Accompanying drawing explanation

Fig. 1 is ground penetrating radar buried target localization method process flow diagram in the present embodiment;

Fig. 2 is Bidimensional Empirical Mode Decomposition algorithm flow chart in the present embodiment;

Fig. 3 is ground penetrating radar actual measurement B-Scan echo in the present embodiment;

Fig. 4 is the image after utilizing Bidimensional Empirical Mode Decomposition to extract first IMF in the present embodiment;

Fig. 5 is the geometry site figure of radar antenna and target B-Scan echo in the present embodiment;

Fig. 6 is that in the present embodiment, matched curve is plotted in the design sketch on original B-Scan image.

Embodiment

Below in conjunction with accompanying drawing, technical scheme of the present invention is described in detail.

As shown in Figure 1, the ground penetrating radar buried target localization method of the present embodiment comprises the steps:

1) two-dimensional empirical mode decomposition is carried out to the detection echo data of ground penetrating radar, obtain two-dimensional empirical modal function component IMF and 1 residual error that K frequency is successively decreased successively;

2) using the eigenwert of the average of front M (M≤K) two-dimensional empirical modal function component as detection echo data;

3) extreme point of the eigenwert of described detection echo data is obtained, as the estimated value of buried target vertex position;

4) velocity of propagation of electromagnetic wave in underground is estimated;

5) according to estimated value and the velocity of propagation of electromagnetic wave in underground of described buried target vertex position, utilize ground penetrating radar hyperbolic curve mathematical model, carry out hyperbolic fit, complete the location of buried target position.

Below above-mentioned steps is described in detail:

Step 1) in ground penetrating radar B-Scan detect echo data carry out two-dimensional empirical mode decomposition, the process of empirical mode decomposition can adopt decomposable process of the prior art, as shown in Figure 2, the two-dimensional empirical mode decomposition process that the present embodiment is preferably as follows:

First Step1 determines the detection echo data I of ground penetrating radar _resall extreme points, concrete employing eight neighborhood method determines I _resthe maximum value that image is all and minimal value;

Step2 is to the detection echo data I of ground penetrating radar _resall extreme points utilize radial basis function to carry out interpolation, the maximum point after interpolation and minimum point use E respectively _iand E _srepresent, after carrying out curve fitting, obtain detection echo data I _resupper and lower envelope;

The concrete form of radial basis function RBF is:

$s\left(x\right)=p_m\left(x\right)+\underset{i=1}{\overset{N}{\&Sigma;}}{\mathrm{\&lambda;}}_i\mathrm{\&Phi;}\left(\right||x-x_i|\left|\right)$

Wherein: s is radial basis function (RBF), p _mpolynomial of lower degree, as linear or secondary or d variable m ^thpolynomial expression, || || represent euclideam norm.λ _ibe RBF coefficient, Φ is real-valued function, the center of the radial basis function RBF that is often known as.

Step3 asks the average E of upper and lower envelope _m=(E _i+ E _s)/2;

Step4 is from original detection echo data I _resin deduct E _m, obtain new detection echo data

Step5 judges according to IMF decision condition whether be an IMF, if an IMF, make first two-dimensional empirical modal function component (IMF) for residual error otherwise, use replace I _res, repeat step a) ~ d) until judge be an IMF, make first two-dimensional empirical modal function component (IMF) for residual error repetition like this, until two-dimensional empirical modal function component IMF and 1 residual error obtaining that K frequency successively decrease successively.

Described IMF decision condition is setting SD threshold value,

$SD=\underset{m=0}{\overset{M}{\&Sigma;}}\underset{n=0}{\overset{N}{\&Sigma;}}\&lsqb;\frac{{|I_{{\mathrm{res}}_{i(j-1)}}(m,n)-I_{{\mathrm{res}}_{ij}}(m,n)|}^2}{I_{{\mathrm{res}}_{i(j-1)}^2}(m,n)}\&rsqb;$

Wherein, with for passing through i-th ^ththe double attenuation results of individual pattern, M, N represent line number and the columns of two dimensional image, represent i-th ^ththe data of the capable n row of m of the jth time decay of individual Mode Decomposition.Preset a threshold value T in reality, stop iteration when SD is less than this threshold value, namely judge an IMF.

Finally obtain two-dimensional empirical modal function component IMF and 1 residual error that K frequency is successively decreased from high to low successively according to the method described above.

For step 2) in the present embodiment preferably before the average of M (M≤K) two-dimensional empirical modal function frequency component as the eigenwert of detection echo data, as shown in Figure 4, this eigenwert can clutter reduction while retaining target location.

Described step 3) in learn that target echo has Hyperbolic Feature according to ground penetrating radar principle, the ordinate on hyperbolic curve summit represents the shortest echo time delay, namely nearest at this measuring point ground penetrating radar distance objective.Therefore, scan by column the eigenwert of the detection echo data chosen, choose minimum value, determine the ordinate on hyperbolic curve summit, hyp horizontal ordinate just represents horizontal level corresponding to target.

For step 4) as shown in Figure 5, by ground penetrating radar principle, obtain ground penetrating radar hyperbolic curve mathematical model:

$\frac{t^2}{t_0^2}-\frac{4{(x-x_0)}^2}{v^2{t}_0^2}=1$

Wherein, x represents aerial position, x ₀represent the horizontal coordinate of representative points position, v represents the velocity of propagation of electromagnetic wave in underground, t ₀expression aerial position is x ₀target reflection echo time delay, t represents that aerial position is the target reflection echo time delay of x.Therefore, apex coordinate (x is obtained ₀, t ₀) and velocity of wave _vget final product accurate localizing objects, having three parameters to need to estimate here is obtain apex coordinate (x respectively ₀, t ₀) and velocity of wave _v.

In above-mentioned steps 3) in estimated representative points coordinate (x ₀, t ₀), below introduce velocity of wave in detail _vestimation process:

A) the minimum value V of a selected velocity of wave _min, utilize frequency-wavenumber migration method to calculate the migration result under given speed value;

B) according to the entropy of image after formulae discovery skew below, E is counted ₁;

$E\left(X\right)=\underset{j=1}{\overset{T}{\&Sigma;}}\left\{\frac{\underset{i=1}{\overset{S}{\&Sigma;}}{x}_{ij}^4}{{\&lsqb;\underset{i=1}{\overset{S}{\&Sigma;}}{x}_{ij}^2\&rsqb;}^2}\right\}$

C) select speed step delta V, use V _min+ Δ V, V _min+ 2 Δ V, V _min+ 3 Δ V ..., to step 2) in processed the detection echo data obtained and carried out calculations of offset, until speed reaches maximum predetermined value V _maxif share n speed parameter, calculate the image entropy after skew, result is designated as E ₂, E ₃..., until E _n.

D) find the velocity amplitude that entropy smallest point is corresponding, this value is the most rational migration velocity parameter v.

The present embodiment preferred aforesaid way estimation velocity of wave v, as other embodiments, estimates in prior art that the mode of velocity of wave v has a lot, introduces no longer in detail here.

For step 5) by step 3) the representative points position that estimates, step 4) speed v that estimates brings in ground penetrating radar hyperbolic curve mathematical model, hyperbola of fit, as shown in Figure 6, completes ground penetrating radar target localization.

Be presented above concrete embodiment, but the present invention is not limited to described embodiment.Basic ideas of the present invention are above-mentioned basic scheme, and for those of ordinary skill in the art, according to instruction of the present invention, designing the model of various distortion, formula, parameter does not need to spend creative work.The change carried out embodiment without departing from the principles and spirit of the present invention, amendment, replacement and modification still fall within the scope of protection of the present invention.

Claims (5)

1. a ground penetrating radar buried target localization method, is characterized in that, comprises the steps:

1) echo data is detected to the B-Scan of ground penetrating radar and carry out two-dimensional empirical mode decomposition, obtain K and 1 residual error;

2) using the eigenwert of the average of front M (M≤K) two-dimensional empirical modal function component as detection echo data;

3) extreme point of the eigenwert of described detection echo data is obtained, as the estimated value of buried target vertex position;

4) velocity of propagation of electromagnetic wave in underground is estimated;

5) according to estimated value and the velocity of propagation of electromagnetic wave in underground of described buried target vertex position, utilize ground penetrating radar hyperbolic curve mathematical model, carry out hyperbolic fit, complete the location of buried target position.

2. a kind of ground penetrating radar buried target localization method according to claim 1, is characterized in that, described step 1) in the detailed process that the detection echo data of ground penetrating radar carries out two-dimensional empirical mode decomposition be:

A) the detection echo data I of ground penetrating radar is first determined _resall extreme points;

B) to the detection echo data I of ground penetrating radar _resall extreme points utilize radial basis function to carry out interpolation, the maximum point after interpolation and minimum point use E respectively _iand E _srepresent, after carrying out curve fitting, obtain detection echo data I _resupper and lower envelope;

C) the average E of upper and lower envelope is asked _m=(E _i+ E _s)/2;

D) from original detection echo data I _resin deduct E _m, obtain new detection echo data $I_{res}^{11}=I_{res}-E_M;$

E) judge according to IMF decision condition whether be an IMF, if an IMF, make first two-dimensional empirical modal function component for residual error otherwise, use replace I _res, repeat step a) ~ d) until judge be an IMF, make first two-dimensional empirical modal function component for residual error repetition like this, until two-dimensional empirical modal function component IMF and 1 residual error obtaining that K frequency successively decrease successively.

3. a kind of ground penetrating radar buried target localization method according to claim 1, it is characterized in that, described step 3) in the obtain manner of estimated value of buried target vertex position be: the eigenwert scanning by column the detection echo data chosen, choose minimum value, determine the ordinate on hyperbolic curve summit, hyp horizontal ordinate just represents horizontal level corresponding to target.

4. a kind of ground penetrating radar buried target localization method according to claim 1, is characterized in that, described step 4) in adopt frequency wave beam deflection method and in conjunction with the velocity of propagation of minimum entropy technology estimation electromagnetic wave in underground.

5. a kind of ground penetrating radar buried target localization method according to claim 1, is characterized in that, described step 5) in ground penetrating radar hyperbolic curve mathematical model be:

$\frac{t^2}{t_0^2}-\frac{4{(x-x_0)}^2}{v^2{t}_0^2}=1$

Wherein, x represents aerial position, x ₀represent the horizontal coordinate of representative points position, v represents the velocity of propagation of electromagnetic wave in underground, t ₀expression aerial position is x ₀target reflection echo time delay, t represents that aerial position is the target reflection echo time delay of x.