CN105182328B - A kind of GPR buried target detection method based on two-dimensional empirical mode decomposition - Google Patents

Abstract

The present invention relates to a kind of GPR buried target detection method based on two-dimensional empirical mode decomposition, specifically include：1) two-dimensional empirical mode decomposition is carried out to the detection echo data of GPR, obtains two-dimensional empirical modal function component IMF and 1 residual error that K frequency is successively decreased successively；2) it regard the average of preceding M (M≤K) two-dimensional empirical modal function components as the characteristic value for detecting echo data；3) extreme point of the detection echo data characteristic value is obtained, the estimate of buried target vertex position is used as；4) spread speed of the estimation electromagnetic wave in underground；5) using GPR hyperbola mathematical modeling, hyperbolic fit is carried out, the positioning of buried target position is completed in the spread speed of underground according to the estimate and electromagnetic wave of the buried target vertex position.The method of the present invention lifts clutter recognition effect while more complete reservation target information, improves the precision of target positioning.

Description

A Ground Penetrating Radar Underground Target Detection Method Based on Two-Dimensional Empirical Mode Decomposition

technical field

The invention relates to the field of ground-penetrating radar detection, in particular to a ground-penetrating radar underground target detection method based on two-dimensional empirical mode decomposition.

Background technique

Ground-penetrating radar is an effective shallow underground target detection technology developed rapidly in recent decades. It is a non-destructive detection method with fast detection speed, high resolution, convenient and flexible operation, and low detection cost. With many advantages, it has been widely used in the detection and positioning of underground targets, such as cavities, pipelines, mines, etc.

The two-dimensional echo data detected by ground penetrating radar is called B-Scan data, which is the data basis for subsequent radar signal processing, target recognition and interpretation, and the ground penetrating radar target positioning technology is also based on B-Scan data. The biggest influence on the accurate positioning of the target is the "clutter" in the ground penetrating radar B-Scan data. Ground penetrating radar clutter can be regarded as various echoes other than target echoes, usually including antenna direct waves, surface echoes, echoes generated by underground heterogeneous media, and echoes generated by false targets, etc. Ground penetrating radar clutter makes it difficult to accurately detect underground targets, especially for shallow buried targets, the target echo is a weaker component compared with the surface echo, and the time between the target echo and the surface echo The delay is very small, and the target echo is easily overwhelmed by clutter such as strong surface echo. Therefore, suppressing clutter in GPR is the primary task to achieve accurate positioning of GPR targets.

The common positioning method is mainly based on the hyperbola extraction of the B-scan image, and the target depth is calculated according to the speed of the extracted hyperbola. Mainly include: the extraction of hyperbola based on neural network requires more data for training, and it is not easy to realize online detection; the pattern recognition method of fuzzy clustering is used for the shallow detection that may exist in both metal pipelines and non-metal pipelines. In other words, it is easy to generate false alarms, and it is easy to miss non-metallic pipeline targets. The method based on image segmentation and Hough transform cannot effectively distinguish between strong clutter and target echo when applied to shallow detection pipelines; the method based on image segmentation and template matching is applied to shallow detection pipelines, due to the The size of the diameter may change, so there are many corresponding templates, resulting in a long algorithm operation time; the curve detection based on morphology is based on the gray value of the image for detection and judgment, and the target area can be judged but it is multi-root Curve, the next step of calculation also needs to process the curve.

Contents of the invention

The invention provides a ground-penetrating radar underground target detection method based on two-dimensional empirical mode decomposition, aiming to solve the problems of complex target positioning methods and low positioning accuracy in the prior art.

In order to solve the problems of the technologies described above, the technical solution of the present invention is:

1) Carry out two-dimensional empirical mode decomposition on the B-Scan detection echo data of the ground penetrating radar, and obtain K two-dimensional empirical mode function components IMF and one residual with decreasing frequencies in turn;

2) The mean value of the first M (M≤K) two-dimensional empirical mode function components is used as the characteristic value of the detection echo data;

3) Obtaining the extreme point of the eigenvalue of the detection echo data as the estimated value of the apex position of the underground target;

4) Estimate the propagation speed of electromagnetic waves in the ground;

5) According to the estimated value of the vertex position of the underground target and the propagation velocity of electromagnetic waves underground, the hyperbolic mathematical model of the ground penetrating radar is used to perform hyperbolic fitting to complete the location of the underground target.

The specific process of carrying out two-dimensional empirical mode decomposition to the detection echo data of ground penetrating radar in described step 1) is:

a) First determine all extreme points of the detection echo data I _res of the ground penetrating radar, and specifically use the eight-neighborhood method to determine all maximum and minimum values of the I _res image;

b) Interpolate all extreme points of the GPR detection echo data I _res using the radial basis function, and the interpolated maximum and minimum points are represented by E _I and E _S respectively, and the curve is fitted After combining, the upper and lower envelopes of the detection echo data I _res are obtained;

The specific form of radial basis function RBF is:

Among them: s is the radial basis function (RBF), p _m is a low-degree polynomial, such as linear or quadratic or m ^th polynomial of d variables, || · || represents the Euclidean norm. λ _i is the RBF coefficient, Φ is a real-valued function, which is often called the center of the radial basis function RBF.

c) Find the mean value of the upper and lower envelopes

E _M = (E _I +E _S )/2; (2)

d) Subtract E _M from the original sounding echo data I _res to obtain new sounding echo data

e) Judgment according to IMF judgment conditions Whether it is an IMF, if it is an IMF, let the first two-dimensional empirical mode function component (IMF) for residual Otherwise, use Instead of I _res , repeat steps a)~d) until it is determined is an IMF, let the first two-dimensional empirical mode function component (IMF) for residual This is repeated until K two-dimensional empirical mode function components IMF with decreasing frequencies and one residual are obtained.

The IMF judgment condition is to set the SD threshold,

in, with is the result of two consecutive attenuations through the i ^th mode, Indicates the value of the m-th row and n-column of the j- ^th attenuation of the ith mode decomposition, and M and N represent the number of rows and columns of the two-dimensional ground-penetrating radar image. In practice, a threshold T is preset, and when the SD is less than the threshold, the iteration is stopped, that is, the judgment is an IMF.

In the step 3), it is known that the target echo has a hyperbolic feature according to the GPR principle, and the ordinate of the apex of the hyperbola represents the shortest echo time delay, that is, the GPR is the closest to the target at this measuring point. Therefore, the eigenvalues of the selected detection echo data are scanned column by column, and the minimum value of the ordinate is selected to determine the ordinate of the apex of the hyperbola. The abscissa of the hyperbola represents the corresponding horizontal position of the target. In the step 4), the frequency beam shifting method combined with the minimum entropy technique is used to estimate the propagation velocity of the electromagnetic wave underground.

The ground penetrating radar hyperbolic mathematical model in the described step 5) is:

Among them, x represents the antenna position, x ₀ represents the horizontal coordinate of the apex position of the target, v represents the propagation velocity of the electromagnetic wave in the ground, t ₀ represents the reflection echo delay of the target at the antenna position x ₀ , and t represents the target at the antenna position x Reflected echo delay.

The ground-penetrating radar underground target detection method based on two-dimensional empirical mode decomposition of the present invention first performs two-dimensional empirical mode decomposition on the detection echo data of the ground-penetrating radar to obtain several single-component signals, and then extracts and detects them according to the single-component signals The characteristic value of the echo data is used to estimate the position of the apex of the target, and then combined with the estimated wave velocity and the principle of the ground penetrating radar, the hyperbola fitting is performed to complete the target positioning. This method improves the clutter suppression effect while retaining the target information more completely, and improves the accuracy of target positioning.

Description of drawings

Fig. 1 is the flow chart of ground penetrating radar underground target location method in the present embodiment;

Fig. 2 is the flow chart of two-dimensional empirical mode decomposition algorithm in the present embodiment;

Fig. 3 is the B-Scan echo image measured by ground penetrating radar in the present embodiment;

Fig. 4 utilizes two-dimensional empirical mode decomposition to extract the image after the first IMF in the present embodiment;

Fig. 5 is the geometric position relationship figure of radar antenna and target B-Scan echo in the present embodiment;

FIG. 6 is an effect diagram of the fitting curve drawn on the original B-Scan image in this embodiment.

detailed description

The technical solution of the present invention will be described in detail below in conjunction with the accompanying drawings.

As shown in Figure 1, the ground penetrating radar underground target detection method based on two-dimensional empirical mode decomposition of the present embodiment includes the following steps:

1) Carry out two-dimensional empirical mode decomposition on the detection echo data of the ground penetrating radar, and obtain K two-dimensional empirical mode function components IMF and one residual with decreasing frequencies in turn;

2) The mean value of the first M (M≤K) two-dimensional empirical mode function components is used as the characteristic value of the detection echo data;

3) Obtaining the extreme point of the eigenvalue of the detection echo data as the estimated value of the apex position of the underground target;

4) Estimate the propagation speed of electromagnetic waves in the ground;

5) According to the estimated value of the vertex position of the underground target and the propagation velocity of electromagnetic waves underground, the hyperbolic mathematical model of the ground penetrating radar is used to perform hyperbolic fitting to complete the location of the underground target.

The above steps are described in detail below:

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

Step1 first determines all the extreme points of the detection echo data I _res of the ground penetrating radar, and specifically adopts the eight-neighborhood method to determine all the maximum and minimum values of the I _res image;

Step2 Interpolate all the extreme points of the ground-penetrating radar detection echo data I _res using the radial basis function, and the interpolated maximum and minimum points are represented by E _I and E _S respectively, and the curve fitting is carried out After obtaining the upper and lower envelopes of the detection echo data I _res ;

The specific form of radial basis function RBF is:

Among them: s is the radial basis function (RBF), p _m is a low-degree polynomial, such as linear or quadratic or m ^th polynomial of d variables, || · || represents the Euclidean norm. λ _i is the RBF coefficient, Φ is a real-valued function, which is often called the center of the radial basis function RBF.

Step3 Calculate the mean value E _M of the upper and lower envelopes = (E _I +E _S )/2;

Step4 Subtract E _M from the original sounding echo data I _res to get new sounding echo data

Step5 Judgment based on IMF judgment conditions Whether it is an IMF, if it is an IMF, let the first two-dimensional empirical mode function component (IMF) for residual Otherwise, use Instead of I _res , repeat steps a)~d) until it is determined is an IMF, let the first two-dimensional empirical mode function component (IMF) for residual This is repeated until K two-dimensional empirical mode function components IMF with decreasing frequencies and one residual are obtained.

The IMF judgment condition is to set the SD threshold,

in, with is the attenuation result of two consecutive times through the i ^th mode, M and N represent the number of rows and columns of the two-dimensional image, Represents the data in the mth row and nth column of the jth decay of the ^ith mode decomposition. In practice, a threshold T is preset, and when the SD is less than the threshold, the iteration is stopped, that is, the judgment is an IMF.

According to the above method, K two-dimensional empirical mode function components IMF with decreasing frequency from high to low and one residual are finally obtained.

For step 2) in this embodiment, the mean value of the first M (M≤K) two-dimensional empirical mode function frequency components is preferred as the feature value of the detection echo data, as shown in Figure 4, while the feature value retains the target position Can suppress clutter.

In the step 3), it is known that the target echo has a hyperbolic feature according to the GPR principle, and the ordinate of the apex of the hyperbola represents the shortest echo time delay, that is, the GPR is the closest to the target at this measuring point. Therefore, the eigenvalues of the selected detection echo data are scanned column by column, and the minimum value is selected to determine the ordinate of the apex of the hyperbola, and the abscissa of the hyperbola represents the corresponding horizontal position of the target.

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

Among them, x represents the antenna position, x ₀ represents the horizontal coordinate of the apex position of the target, v represents the propagation velocity of the electromagnetic wave in the ground, t ₀ represents the reflection echo delay of the target at the antenna position x ₀ , and t represents the target at the antenna position x Reflected echo delay. Therefore, the target can be accurately located by obtaining the vertex coordinates (x ₀ , t ₀ ) and the wave velocity v. Here, there are three parameters to be estimated, which are to obtain the vertex coordinates (x ₀ , t ₀ ) and the wave velocity v.

The target vertex coordinates (x ₀ ,t ₀ ) have been estimated in the above step 3), and the estimation process of the wave velocity v is introduced in detail below:

a) Select a minimum value V _min of the wave velocity, and use the frequency wavenumber migration method to calculate the migration result at a given velocity value;

b) Calculate the entropy of the shifted image according to the following formula, count as E ₁ ;

c) Select the speed step size ΔV, use V _min + ΔV, V _min + 2ΔV, V _min + 3ΔV, ..., and perform offset calculation on the probe echo data processed in step 2) until the speed reaches the maximum predetermined value Value V _max , assuming that n speed parameters are shared, the image entropy after migration is calculated, and the results are recorded as E ₂ , E ₃ ,..., until E _n .

d) Find the velocity value corresponding to the minimum entropy point, which is the most reasonable offset velocity parameter v.

In this embodiment, the above method is preferred for estimating the wave velocity v. As other implementation manners, there are many methods for estimating the wave velocity v in the prior art, which will not be described in detail here.

For step 5) bring the target vertex position estimated in step 3) and the velocity v estimated in step 4) into the ground-penetrating radar hyperbolic mathematical model, and fit the hyperbola, as shown in Figure 6, to complete the ground-penetrating radar target position.

Specific embodiments have been given above, but the present invention is not limited to the described embodiments. The basic idea of the present invention lies in the above-mentioned basic scheme. For those of ordinary skill in the art, according to the teaching of the present invention, it does not need to spend creative labor to design various deformation models, formulas, and parameters. Changes, modifications, substitutions and variations to the implementations without departing from the principle and spirit of the present invention still fall within the protection scope of the present invention.

Claims (5)

1. A ground penetrating radar underground target detection method based on two-dimensional empirical mode decomposition is characterized by comprising the following steps:

1) performing two-dimensional empirical mode decomposition on B-Scan detection echo data of the ground penetrating radar to obtain K two-dimensional empirical mode function components IMF with sequentially decreasing frequency and 1 residual error;

2) taking the mean value of the first M (M is less than or equal to K) two-dimensional empirical mode function components as the characteristic value of the detection echo data;

3) acquiring an extreme point of the characteristic value of the detection echo data as an estimated value of the vertex position of the underground target;

4) estimating the propagation speed of the electromagnetic wave in the underground;

5) and according to the estimated value of the top position of the underground target and the propagation speed of the electromagnetic wave in the underground, performing hyperbolic fitting by using a hyperbolic mathematical model of the ground penetrating radar to complete positioning of the underground target position.

2. The method for detecting the underground target of the ground penetrating radar based on the two-dimensional empirical mode decomposition according to claim 1, wherein the specific process of performing the two-dimensional empirical mode decomposition on the detection echo data of the ground penetrating radar in the step 1) is as follows:

a) firstly, determining detection echo data I of ground penetrating radar_resAll extreme points of (a);

b) detection echo data I of ground penetrating radar_resAll the extreme points are interpolated by using the radial basis function, and the interpolated maximum point and minimum point are respectively interpolated by using E_IAnd E_SRepresenting, obtaining detection echo data I after curve fitting_resUpper and lower envelopes of;

c) calculating the average value E of the upper and lower envelopes_M＝(E_I+E_S)/2；

d) From raw probe echo data I_resMinus E_MObtaining new detection echo data

e) Determining condition judgment according to two-dimensional empirical mode function (IMF)Whether the two-dimensional empirical mode function component IMF is determined, if the two-dimensional empirical mode function component IMF is determined, the first two-dimensional empirical mode function component IMF is determinedIs composed ofResidual errorOtherwise, useIn place of I_resRepeating steps a) to d) until a decision is madeFor a two-dimensional empirical mode function component IMF, let the first two-dimensional empirical mode function component IMFIs composed ofResidual errorAnd repeating the steps until K two-dimensional empirical mode function components IMF with sequentially decreasing frequency and 1 residual error are obtained.

3. The method for detecting the underground target of the ground penetrating radar based on the two-dimensional empirical mode decomposition according to claim 1, wherein the estimation value of the vertex position of the underground target in the step 3) is obtained by: and scanning the selected characteristic values of the detected echo data row by row, selecting the minimum value, determining the ordinate of the vertex of the hyperbola, and representing the horizontal position corresponding to the target by the abscissa of the hyperbola.

4. The method for detecting the underground target of the ground penetrating radar based on the two-dimensional empirical mode decomposition according to claim 1, wherein the propagation speed of the electromagnetic wave in the underground is estimated in the step 4) by adopting a frequency beam shifting method and combining a minimum entropy technology.

5. The method for detecting the underground target of the ground penetrating radar based on the two-dimensional empirical mode decomposition according to claim 1, wherein the hyperbolic mathematical model of the ground penetrating radar in the step 5) is as follows:

<mrow> <mfrac> <msup> <mi>t</mi> <mn>2</mn> </msup> <msubsup> <mi>t</mi> <mn>0</mn> <mn>2</mn> </msubsup> </mfrac> <mo>-</mo> <mfrac> <mrow> <mn>4</mn> <msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>-</mo> <msub> <mi>x</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> <mrow> <msup> <mi>v</mi> <mn>2</mn> </msup> <msubsup> <mi>t</mi> <mn>0</mn> <mn>2</mn> </msubsup> </mrow> </mfrac> <mo>=</mo> <mn>1</mn> </mrow>

where x denotes the antenna position, x₀Horizontal coordinates representing the position of the apex of the target, v represents the propagation velocity of the electromagnetic wave in the ground, t₀Indicates the position of the antenna as x₀T represents the target reflection echo time delay with the antenna position x.