Through-wall radar imaging method, system, terminal device and readable storage medium

Abstract

The present invention provides a through-wall radar imaging method, system, terminal device and readable storage medium, the method comprising: obtaining radar echo data in a preset radar detection scene, discretizing the imaging scene in the preset radar detection scene; constructing a dictionary matrix according to the discretization result of the imaging scene, and constructing a low-rank sparse signal model according to the dictionary matrix; determining a low-rank sparse optimization problem in the low-rank sparse signal model, and constructing a through-wall radar imaging model according to the determined low-rank sparse optimization problem; training the through-wall radar imaging model until the through-wall radar imaging model converges, inputting the echo data to be imaged into the converged through-wall radar imaging model for imaging analysis, and obtaining a reconstructed image. The present invention transforms the clutter suppression of the wall and the reconstruction of the target image into solving the low-rank sparse optimization problem, effectively suppresses the reflected clutter of the wall, and improves the imaging quality of the reconstructed image.

Description

Technical Field

[0001] The present invention relates to the field of radar imaging, and in particular to a through-wall radar imaging method, system, terminal device and readable storage medium. Background Art

[0002] Through-the-wall radar imaging is a sensing technology that can be used by responders to see obscuring obstacles in search and rescue missions, or to deploy law enforcement and military personnel to maintain situational awareness during tactical operations. Through-the-wall radar uses the ability of electromagnetic waves to penetrate non-metallic building materials, providing a unique opportunity for imaging targets behind walls or in closed environments. Through-the-wall radar imaging is widely used in military and civilian fields such as urban street fighting, disaster relief and housing security. It can detect, locate, track and image targets, thereby being able to detect hidden targets behind walls.

[0003] In through-wall radar imaging, due to the presence of obstacles such as non-transparent walls between the target and the radar antenna, the echo signal received by the antenna contains a large amount of wall reflection clutter, which seriously affects the imaging quality and even makes the target invisible. In order to facilitate the analysis of the performance of the imaging algorithm, some traditional imaging algorithms do not consider the wall reflection clutter or assume that it has been filtered out by the background subtraction method before imaging processing. However, in actual situations, it is almost difficult to obtain data of empty scenes, which makes the background subtraction method unusable.

[0004] In the existing radar imaging process, the reflection clutter of the wall in the radar detection scene cannot be effectively removed, resulting in low quality of the reconstructed image after imaging. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a through-wall radar imaging method, system, terminal device and readable storage medium to solve the problem in the prior art that the reflected clutter of the wall in the radar detection scene cannot be effectively removed, resulting in poor quality of the reconstructed image after imaging.

[0006] A first aspect of an embodiment of the present invention provides a through-wall radar imaging method, comprising:

[0007] Acquire radar echo data in a preset radar detection scene, and discretize the imaging scene in the preset radar detection scene;

[0008] Constructing a dictionary matrix according to the discretization result of the imaging scene, and constructing a low-rank sparse signal model according to the dictionary matrix, wherein the low-rank sparse signal model is used to characterize the low-rank characteristics of the wall clutter and the spatial sparse characteristics of the target behind the wall to be detected in the preset radar detection scene;

[0009] Determine a low-rank sparse optimization problem in the low-rank sparse signal model, and construct a through-wall radar imaging model according to the determined low-rank sparse optimization problem;

[0010] The through-wall radar imaging model is trained until the through-wall radar imaging model converges, and the echo data to be imaged is input into the converged through-wall radar imaging model for imaging analysis to obtain a reconstructed image.

[0011] Further, the obtaining of radar echo data in a preset radar detection scene includes:

[0012] A single-station stepped frequency system is adopted, and the co-located transmitting and receiving antenna is stepped forward from left to right in a direction parallel to the wall, and a first preset number of measuring positions are set in the direction of the measuring line, and each of the measuring positions transmits a second preset number of frequency point signals;

[0013] The radar echo data measured at the nth frequency signal and the mth measurement position is z n,m ;

[0014]

[0015] is the wall reflection signal in the preset radar detection scene, is the target reflection signal in the preset radar detection scene, v n，m A measurement noise signal in the preset radar detection scene;

[0016] The wall reflects the signal for:

[0017]

[0018] The target reflected signal for:

[0019]

[0020] Among them, σ w is the reflection coefficient of the wall in the preset radar detection scene, f n is the mth frequency signal, τ w represents the round-trip transmission delay from the mth measurement position to the wall, which is independent of the measurement position, p is the total number of targets, σ p is the reflection coefficient of the pth target, τ m，q represents the round-trip delay between the mth measurement location and the qth target;

[0021] The total echo signal in the radar echo data is:

[0022] Z=Z w+Z t +V.

[0023] Furthermore, the discretizing the imaging scene in the preset radar detection scene includes:

[0024] The imaging scene is two-dimensionally discretized to obtain discrete grids, and the total number of discrete grids Q=KxL′, wherein K and L′ are the numbers of discrete grids corresponding to the range direction and the azimuth direction in the imaging scene, respectively.

[0025] Furthermore, constructing a dictionary matrix according to the discretization result of the imaging scene includes:

[0026] Performing a column stacking operation on the discrete grids to obtain a scene reflectivity vector;

[0027] The qth element S in the scene reflectivity vector q for:

[0028]

[0029] The target reflection signal is:

[0030] Z t =Ψ1S

[0031] Ψ1 is a dictionary matrix of MN×Q dimensions:

[0032] Ψ1=exp(-j2πf n τ m，p )

[0033] q=1,2,...,Q,f n is the nth frequency signal, p is the total number of targets, τ m，q represents the round-trip delay between the mth measurement location and the qth target;

[0034] S is a sparse vector, the qth element of the reflectivity vector L of the wall q It is expressed as:

[0035]

[0036] The wall reflection signal is:

[0037] Z w =Ψ2L

[0038] Ψ2 is a dictionary matrix of MN×Q dimensions:

[0039] Ψ2=exp(-j2πf n τ w )

[0040] τ wrepresents the round-trip transmission delay from the mth measurement location to the wall, and L is a low-rank matrix with wall reflection.

[0041] Furthermore, the low-rank sparse signal model constructed according to the dictionary matrix is:

[0042] D=Ψ1L+Ψ2S+V

[0043] D is the observation matrix, L is the low-rank matrix with wall reflection, S is the sparse matrix with target reflection, N is the measurement noise matrix, Ψ1 and Ψ2 are the dictionary matrices.

[0044] Furthermore, the low-rank sparse optimization problem is determined to be formula (1):

[0045]

[0046] ||L|| * represents the nuclear norm, which is the sum of the singular values ​​of L, ||S|| 1，2 It is mixed 1，2 norm, whose value is equal to l in each row of S 1，2 norms and, λ1 and λ2 are tuning parameters used to balance the nuclear norm and the norm.

[0047] Furthermore, constructing a through-wall radar imaging model according to the determined low-rank sparse optimization problem includes:

[0048] definition: A = [Ψ1, Ψ2];

[0049] I is the identity matrix, and the low-rank sparse optimization problem is described as formula (2):

[0050]

[0051] in, ρ1=||·|| * ,ρ2=||·|| 1，2 ;

[0052]

[0053] P i Represents matrices P1 and P 1，2 , prox h (X) is separated in L and S, and formula (3) is obtained:

[0054]

[0055] is the soft threshold operator;

[0056] SVTα (X) = Udiag(max(0,σ i -α))V H ,i=1,...,r,is the singular value threshold operator;

[0057] When X=U∑V H ,∑=diag(σ i , ..., σ r ) is the diagonal matrix of the eigenvalues ​​of X. Applying the proximal mapping of formula (3) to the quadratic partial gradient of formula (2) yields:

[0058]

[0059] Ψ1 H and Ψ2 H They are respectively expressed as the conjugate forms of Ψ1 and Ψ2, where Ψ1 and Ψ2 are dictionary matrices, and the minimum value of formula (1) is obtained:

[0060]

[0061] Or it can be expressed as formula (4):

[0062]

[0063] L f is the Lipschitz constant of the quadratic term in formula (2), that is, A H The spectral norm of A, the kth iteration is used as the kth layer of the feedforward network in the through-wall radar imaging model, and the through-wall radar imaging model is formed according to formula (4), replacing each matrix that depends on Ψ1 and Ψ2 with a convolutional layer of corresponding size The equation for the kth layer is:

[0064]

[0065] Wherein, * represents the corresponding convolution operator in the through-wall radar imaging model, for The singular value threshold operator of for The soft threshold operator.

[0066] A second aspect of an embodiment of the present invention provides a through-wall radar imaging system, including:

[0067] A radar echo data acquisition unit, used to acquire radar echo data in a preset radar detection scene, and discretize an imaging scene in the preset radar detection scene;

[0068] A signal model construction unit, used to construct a dictionary matrix according to the discretization result of the imaging scene, and to construct a low-rank sparse signal model according to the dictionary matrix, wherein the low-rank sparse signal model is used to characterize the low-rank characteristics of the wall clutter and the spatial sparse characteristics of the target behind the wall to be detected in the preset radar detection scene;

[0069] A low-rank sparse problem determination unit, used to determine a low-rank sparse optimization problem in the low-rank sparse signal model, and construct a through-wall radar imaging model according to the determined low-rank sparse optimization problem;

[0070] The imaging analysis unit is used to perform model training on the through-wall radar imaging model until the through-wall radar imaging model converges, and input the echo data to be imaged into the converged through-wall radar imaging model for imaging analysis to obtain a reconstructed image.

[0071] A third aspect of an embodiment of the present invention provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the terminal device, wherein when the processor executes the computer program, each step of the through-wall radar imaging method provided by the first solution is implemented.

[0072] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the through-wall radar imaging method provided by the first solution are implemented.

[0073] A through-wall radar imaging method, system, terminal device and readable storage medium provided by the embodiments of the present invention have the following beneficial effects: utilizing the low-rank characteristics of wall clutter and the spatial sparse characteristics of the target behind the wall to be measured, the clutter suppression of the wall and the target image reconstruction problem are transformed into a low-rank sparse optimization problem, which effectively suppresses the reflected clutter of the wall and improves the imaging quality of the reconstructed image. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0075] Figure 1 is a flow chart of an implementation of a through-wall radar imaging method provided by an embodiment of the present invention;

[0076] Figure 2 yes Figure 1 A schematic diagram of the structure of a preset radar detection scenario provided in an embodiment;

[0077] Figure 3 yes Figure 1 An architectural diagram of a single-layer network in a through-wall radar imaging model provided in an embodiment;

[0078] Figure 4 yes Figure 1 Label map of radar detection scene provided by the embodiment;

[0079] Figure 5 yes Figure 1 A reconstructed image output by a converged through-wall radar imaging model provided in an embodiment;

[0080] Figure 6 is a flow chart of an implementation of a through-wall radar imaging method provided by another embodiment of the present invention;

[0081] Figure 7 is a structural block diagram of a through-wall radar imaging system provided by an embodiment of the present invention;

[0082] Figure 8 It is a structural block diagram of a terminal device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0083] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0084] The through-wall radar imaging method involved in the embodiment of the present invention can be executed by a control device or terminal (hereinafter referred to as a "mobile terminal").

[0085] See also Figure 1 , Figure 1 A flowchart of an implementation of a through-wall radar imaging method provided by an embodiment of the present invention is shown, including:

[0086] Step S10, acquiring radar echo data in a preset radar detection scene, and discretizing an imaging scene in the preset radar detection scene;

[0087] Optionally, in this step, obtaining radar echo data in a preset radar detection scene includes:

[0088] A single-station stepped frequency system is adopted, and the co-located transmitting and receiving antenna is stepped forward from left to right in a direction parallel to the wall, and a first preset number of measuring positions are set in the direction of the measuring line, and each of the measuring positions transmits a second preset number of frequency point signals;

[0089] Among them, see Figure 2In this embodiment, the number of measurement positions is 43, the number of frequency point signals transmitted at each measurement position is 64, the imaging scene range is 1m×1m, the spatial resolution is 0.02m, the center frequency is 1.5Ghz, the bandwidth is 1Ghz, and the antenna scanning step is 0.02m.

[0090] The radar echo data measured at the nth frequency signal and the mth measurement position is Z n,m ;

[0091]

[0092] is the wall reflection signal in the preset radar detection scene, is the target reflection signal in the preset radar detection scene, v n,m A measurement noise signal in the preset radar detection scene;

[0093] The wall reflects the signal for:

[0094]

[0095] The target reflected signal for:

[0096]

[0097] Among them, σ w is the reflection coefficient of the wall in the preset radar detection scene, f n is the mth frequency signal, τ w represents the round-trip transmission delay from the mth measurement position to the wall, which is independent of the measurement position, p is the total number of targets, σ p is the reflection coefficient of the pth target, τ m，q represents the round-trip delay between the mth measurement location and the qth target;

[0098] The total echo signal in the radar echo data is:

[0099] Z=Z w +Z t +V.

[0100] Furthermore, in this step, the discretization of the imaging scene in the preset radar detection scene includes:

[0101] The imaging scene is two-dimensionally discretized to obtain discrete grids, and the total number of discrete grids Q=KxL′, wherein K and L′ are the numbers of discrete grids corresponding to the range direction and the azimuth direction in the imaging scene, respectively.

[0102] Step S20, constructing a dictionary matrix according to the discretization result of the imaging scene, and constructing a low-rank sparse signal model according to the dictionary matrix;

[0103] The low-rank sparse signal model is used to characterize the low-rank characteristics of wall clutter and the spatial sparse characteristics of the target behind the wall to be detected in the preset radar detection scene;

[0104] Optionally, in this step, constructing a dictionary matrix according to the discretization result of the imaging scene includes:

[0105] Performing a column stacking operation on the discrete grids to obtain a scene reflectivity vector;

[0106] The qth element S in the scene reflectivity vector q for:

[0107]

[0108] The target reflection signal is:

[0109] Z t =Ψ1S

[0110] Ψ1 is a dictionary matrix of MN×Q dimensions:

[0111] Ψ1=exp(-j2πf n τ m，p )

[0112] q=1,2,...,Q,f n is the nth frequency signal, p is the total number of targets, τ m，q represents the round-trip delay between the mth measurement location and the qth target;

[0113] S is a sparse vector, the qth element of the reflectivity vector L of the wall q It is expressed as:

[0114]

[0115] The wall reflection signal is:

[0116] Z w =Ψ2L

[0117] Ψ2 is a dictionary matrix of MN×Q dimensions:

[0118] Ψ2=exp(-j2πf n τ w )

[0119] τ w represents the round-trip transmission delay from the mth measurement location to the wall, and L is a low-rank matrix with wall reflection.

[0120] Furthermore, in this step, the low-rank sparse signal model constructed according to the dictionary matrix is:

[0121] D=Ψ1L+Ψ2S+V

[0122] D is the observation matrix, L is the low-rank matrix with wall reflection, S is the sparse matrix with target reflection, N is the measurement noise matrix, Ψ1 and Ψ2 are the dictionary matrices.

[0123] In this embodiment, the imaging scene is divided into 51*51 grids at equal intervals of 0.02m, that is, K=L=51, Q=2601. Therefore, the dimension of the dictionary matrix constructed according to the parameters of the antenna and frequency point signals and the spatial discrete grid is MN×Q=2752×2601.

[0124] Step S30, determining a low-rank sparse optimization problem in the low-rank sparse signal model, and constructing a through-wall radar imaging model according to the determined low-rank sparse optimization problem;

[0125] Optionally, in this step, the low-rank sparse optimization problem determined is formula (1):

[0126]

[0127] ‖L‖ * represents the nuclear norm, which is the sum of L singular values, ‖S‖ 1,2 It is mixed 1，2 norm, whose value is equal to l in each row of S 1，2 norms and, λ1 and λ2 are tuning parameters used to balance the nuclear norm and the norm.

[0128] Furthermore, in this step, constructing a through-wall radar imaging model according to the determined low-rank sparse optimization problem includes:

[0129] definition: A = [Ψ1, Ψ2];

[0130] I is the identity matrix, and the low-rank sparse optimization problem is described as formula (2):

[0131]

[0132] in, ρ1=||·|| * ,ρ2=‖·‖ 1，2 ;

[0133]

[0134] P i Represents matrices P1 and P1,2, and proxh (x) is separated in L and S, and formula (3) is obtained:

[0135]

[0136] is the soft threshold operator;

[0137] SVT α (X) = Udiag(max(0,σ i -α))V H , i = 1, ..., r, is the singular value threshold operator;

[0138] When X=U∑V H ,∑=diag(σ i , ..., σ r ) is the diagonal matrix of the eigenvalues ​​of X. Applying the proximal mapping of formula (3) to the quadratic partial gradient of formula (2) yields:

[0139]

[0140] Ψ1 H and Ψ2 H They are respectively expressed as the conjugate forms of Ψ1 and Ψ2, where Ψ1 and Ψ2 are dictionary matrices, and the minimum value of formula (1) is obtained:

[0141]

[0142] Or it can be expressed as formula (4):

[0143]

[0144] L f is the Lipschitz constant of the quadratic term in formula (2), that is, A H The spectral norm of A, the kth iteration is used as the kth layer of the feedforward network in the through-wall radar imaging model, and the through-wall radar imaging model is formed according to formula (4), replacing each matrix that depends on Ψ1 and Ψ2 with a convolutional layer of corresponding size The equation for the kth layer is:

[0145]

[0146] Wherein, * represents the corresponding convolution operator in the through-wall radar imaging model, for The singular value threshold operator of for The soft threshold operator.

[0147] See also Figure 3, is the architecture diagram of a single-layer network in the through-wall radar imaging model. The through-wall radar imaging model is a deep neural network model. The single-layer network in the through-wall radar imaging model has 10 layers. The convolution kernel size of the first three layers is 5*5*1, the sliding step (strides) is (1,1,1), and the inner margin (padding) is (2,2,0). The convolution kernel size of the last seven layers is 3*3*1, the strides are (1,1,1), and the padding is (1,1,0).

[0148] Step S40, training the through-wall radar imaging model until the through-wall radar imaging model converges, and inputting the echo data to be imaged into the converged through-wall radar imaging model for imaging analysis to obtain a reconstructed image;

[0149] Among them, see Figures 4 to 5 , Figure 4 is the label map of the radar detection scene, Figure 5 It is the reconstructed image output by the converged through-wall radar imaging model. The converged through-wall radar imaging model can effectively suppress the strong reflection clutter of the wall and obtain a high-precision reconstructed image of the radar detection target behind the wall.

[0150] In this embodiment, the low-rank characteristics of wall clutter and the spatial sparse characteristics of the target behind the wall to be measured are used to transform the clutter suppression of the wall and the target image reconstruction problem into a low-rank sparse optimization problem, which effectively suppresses the reflected clutter of the wall and improves the imaging quality of the reconstructed image. Based on the through-wall radar imaging model, the original low-rank sparse regularization problem can be solidified into the neural network structure. The through-wall radar imaging model is expanded into a deep network through an iterative algorithm to achieve image reconstruction. The through-wall radar imaging model uses a convolutional layer instead of a fully connected layer, which can achieve better performance than traditional iterative versions and other deep learning architectures.

[0151] See also Figure 6 , Figure 6 FIG. 1 is a flowchart of a through-wall radar imaging method according to another embodiment of the present invention. Figure 1 Embodiment, the through-wall radar imaging method provided in this embodiment is used for Figure 1 Step S40 in the embodiment is further refined to include:

[0152] Step S41, obtaining training samples, and generating a test set according to the training samples;

[0153] Among them, radar echo data and tag data are included;

[0154] Step S42, using mean square error as the loss function, setting the learning rate to 0.001 and the batch size to 5, setting the number of training iterations to 3500, and performing model training on the through-wall radar imaging model.

[0155] Step S43: if it is detected that the number of iterations of the through-wall radar imaging model is greater than the number threshold, or the model loss value of the through-wall radar imaging model is less than the loss threshold, it is determined that the through-wall radar imaging model has converged;

[0156] In this embodiment, both the number threshold and the loss threshold can be set according to needs. If it is detected that the iteration number of the through-wall radar imaging model is greater than the number threshold, or the model loss value of the through-wall radar imaging model is less than the loss threshold, then the through-wall radar imaging model after the current iteration is determined to be converged. Based on the converged through-wall radar imaging model, the input echo data to be imaged can be effectively analyzed to obtain a reconstructed image corresponding to the echo data to be imaged.

[0157] See also Figure 7 , Figure 7 1 is a structural block diagram of a through-wall radar imaging system 100 provided in an embodiment of the present invention. In this embodiment, each unit included in the through-wall radar imaging system 100 is used to execute Figure 1 , Figure 6 For details, please refer to the steps in the corresponding embodiment. Figure 1 , Figure 6 as well as Figure 1 , Figure 6 For the convenience of explanation, only the parts related to this embodiment are shown. Figure 7 The through-wall radar imaging system 100 includes: a radar echo data acquisition unit 10, a signal model construction unit 11, a low-rank sparse problem determination unit 12 and an imaging analysis unit 13, wherein:

[0158] The radar echo data acquisition unit 10 is used to acquire radar echo data in a preset radar detection scene and discretize the imaging scene in the preset radar detection scene.

[0159] The radar echo data acquisition unit 10 is further used to: adopt a single-station step frequency system, step the co-located transmitting and receiving antennas from left to right in a direction parallel to the wall, and set a first preset number of measurement positions in the measurement line direction, and each of the measurement positions transmits a second preset number of frequency point signals;

[0160] The radar echo data measured at the nth frequency signal and the mth measurement position is Z n，m ;

[0161]

[0162] is the wall reflection signal in the preset radar detection scene, is the target reflection signal in the preset radar detection scene, V n，m A measurement noise signal in the preset radar detection scene;

[0163] The wall reflects the signal for:

[0164]

[0165] The target reflected signal for:

[0166]

[0167] Among them, σ w is the reflection coefficient of the wall in the preset radar detection scene, f n is the mth frequency signal, τ w represents the round-trip transmission delay from the mth measurement position to the wall, which is independent of the measurement position, p is the total number of targets, σ p is the reflection coefficient of the pth target, τ m，q represents the round-trip delay between the mth measurement location and the qth target;

[0168] The total echo signal in the radar echo data is:

[0169] Z=Z w +Z t +V.

[0170] Optionally, the radar callback data acquisition unit 10 is further used to: discretize the imaging scene in two dimensions to obtain discrete grids, the total number of discrete grids Q=KxL′, wherein K and L′ are the numbers of discrete grids corresponding to the range and azimuth directions in the imaging scene, respectively.

[0171] The signal model construction unit 11 is used to construct a dictionary matrix according to the discretization result of the imaging scene, and to construct a low-rank sparse signal model according to the dictionary matrix. The low-rank sparse signal model is used to characterize the low-rank characteristics of the wall clutter and the spatial sparse characteristics of the target behind the wall to be detected in the preset radar detection scene.

[0172] The signal model building unit 11 is further used to: perform a column stacking operation on the discrete grids to obtain a scene reflectivity vector;

[0173] The qth element S in the scene reflectivity vector q for:

[0174]

[0175] The target reflection signal is:

[0176] Z t =Ψ1S

[0177] Ψ1 is a dictionary matrix of MN×Q dimensions:

[0178] Ψ1=exp(-j2πf n τ m，p )

[0179] q=1,2,...,Q,f n is the nth frequency signal, p is the total number of targets, τ m，q represents the round-trip delay between the mth measurement location and the qth target;

[0180] S is a sparse vector, the qth element of the reflectivity vector L of the wall q It is expressed as:

[0181]

[0182] The wall reflection signal is:

[0183] Z w =Ψ2L

[0184] Ψ2 is a dictionary matrix of MN×Q dimensions:

[0185] Ψ2=exp(-j2πf n τ w )

[0186] τ w represents the round-trip transmission delay from the mth measurement location to the wall, and L is a low-rank matrix with wall reflection.

[0187] Optionally, in this embodiment, the low-rank sparse signal model constructed according to the dictionary matrix is:

[0188] D=Ψ1L+Ψ2S+V

[0189] D is the observation matrix, L is the low-rank matrix with wall reflection, S is the sparse matrix with target reflection, N is the measurement noise matrix, Ψ1 and Ψ2 are the dictionary matrices.

[0190] The low-rank sparse problem determination unit 12 is used to determine the low-rank sparse optimization problem in the low-rank sparse signal model, and construct a through-wall radar imaging model according to the determined low-rank sparse optimization problem.

[0191] The low-rank sparse optimization problem determined is formula (1):

[0192]

[0193] ||L|| * represents the nuclear norm, which is the sum of the singular values ​​of L, ||S|| 1，2 It is mixed 1，2 norm, whose value is equal to l in each row of S 1，2 norms and, λ1 and λ2 are tuning parameters used to balance the nuclear norm and the norm.

[0194] Optionally, in this step, the low-rank sparse problem determining unit 12 is further used to:

[0195] definition: A = [Ψ1, Ψ2];

[0196] I is the identity matrix, and the low-rank sparse optimization problem is described as formula (2):

[0197]

[0198] in, ρ1=||·|| * ,ρ2=||·|| 1，2 ;

[0199]

[0200] P i Represents matrices P1 and P 1，2 , prox h (X) is separated in L and S, and formula (3) is obtained:

[0201]

[0202] is the soft threshold operator;

[0203] SVT α (X) = Udiag(max(0 , σ i -α))V H , i = 1, ..., r, is the singular value threshold operator;

[0204] When X=U∑V H ,∑=diag(σ i , ..., σ r ) is the diagonal matrix of the eigenvalues ​​of X. Applying the proximal mapping of formula (3) to the quadratic partial gradient of formula (2) yields:

[0205]

[0206] Ψ1H and Ψ2 H They are respectively expressed as the conjugate forms of Ψ1 and Ψ2, where Ψ1 and Ψ2 are dictionary matrices, and the minimum value of formula (1) is obtained:

[0207]

[0208] Or it can be expressed as formula (4):

[0209]

[0210] L f is the Lipschitz constant of the quadratic term in formula (2), that is, A H The spectral norm of A, the kth iteration is used as the kth layer of the feedforward network in the through-wall radar imaging model, and the through-wall radar imaging model is formed according to formula (4), replacing each matrix that depends on Ψ1 and Ψ2 with a convolutional layer of corresponding size The equation for the kth layer is:

[0211]

[0212] Wherein, * represents the corresponding convolution operator in the through-wall radar imaging model, for The singular value threshold operator of for The soft threshold operator.

[0213] The imaging analysis unit 13 is used to perform model training on the through-wall radar imaging model until the through-wall radar imaging model converges, and input the echo data to be imaged into the converged through-wall radar imaging model for imaging analysis to obtain a reconstructed image.

[0214] In this embodiment, the low-rank characteristics of wall clutter and the spatial sparse characteristics of the target behind the wall to be measured are used to transform the clutter suppression of the wall and the target image reconstruction problem into a low-rank sparse optimization problem, which effectively suppresses the reflected clutter of the wall and improves the imaging quality of the reconstructed image.

[0215] Figure 8 FIG. 2 is a block diagram of a terminal device 2 provided in another embodiment of the present invention. Figure 8 As shown, the terminal device 2 of this embodiment includes: a processor 20, a memory 21, and a computer program 22 stored in the memory 21 and executable on the processor 20, such as a program of a through-wall radar imaging method. When the processor 20 executes the computer program 22, the steps in each embodiment of the through-wall radar imaging method described above are implemented, such as Figure 1 S10 to S40 as shown, or Figure 6Alternatively, the processor 20 implements the above when executing the computer program 22 Figure 7 The functions of each unit in the corresponding embodiment are, for example, Figure 7 For details on the functions of units 10 to 13, please refer to Figure 7 The relevant descriptions in the corresponding embodiments are not repeated here.

[0216] Exemplarily, the computer program 22 may be divided into one or more units, which are stored in the memory 21 and executed by the processor 20 to complete the present invention. The one or more units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program 22 in the terminal device 2. For example, the computer program 22 may be divided into a road identification unit 10, a first type marking unit 11, a second type marking unit 12, and a road type determination unit 13, and the specific functions of each unit are as described above.

[0217] The terminal device may include, but is not limited to, a processor 20 and a memory 21. Those skilled in the art will appreciate that Figure 8 It is only an example of the terminal device 2 and does not constitute a limitation of the terminal device 2. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal device may also include input and output devices, network access devices, buses, etc.

[0218] The processor 20 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0219] The memory 21 may be an internal storage unit of the terminal device 2, such as a hard disk or memory of the terminal device 2. The memory 21 may also be an external storage device of the terminal device 2, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device 2. Further, the memory 21 may also include both an internal storage unit and an external storage device of the terminal device 2. The memory 21 is used to store the computer program and other programs and data required by the terminal device. The memory 21 may also be used to temporarily store data that has been output or is to be output.

[0220] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0221] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A through-wall radar imaging method, characterized in that: include: Acquire radar echo data in a preset radar detection scene, and discretize the imaging scene in the preset radar detection scene; Constructing a dictionary matrix according to the discretization result of the imaging scene, and constructing a low-rank sparse signal model according to the dictionary matrix, wherein the low-rank sparse signal model is used to characterize the low-rank characteristics of the wall clutter and the spatial sparse characteristics of the target behind the wall to be detected in the preset radar detection scene; Determine a low-rank sparse optimization problem in the low-rank sparse signal model, and construct a through-wall radar imaging model according to the determined low-rank sparse optimization problem; Performing model training on the through-wall radar imaging model until the through-wall radar imaging model converges, and inputting the echo data to be imaged into the converged through-wall radar imaging model for imaging analysis to obtain a reconstructed image; Wherein, constructing a through-wall radar imaging model according to the determined low-rank sparse optimization problem includes: definition: A = [Ψ1, Ψ2]; The low-rank sparse optimization problem is determined as formula (1): D is the observation matrix, ‖L‖ * represents the nuclear norm, which is the sum of L singular values, ‖S‖ 1,2 It is mixed 1，2 norm, whose value is equal to l in each row of S 1，2 norm and; The low-rank sparse optimization problem is described as formula (2): in, ρ1=||·|| * ,ρ2=||·|| 1，2 ; Prox h (X) is separated in L and S, and formula (3) is obtained: is the soft threshold operator; SVT α (X) = Udiag(max(0,σ b -α))V H , b = 1, ..., r, is the singular value threshold operator; When X=U∑V H ,∑=diag(σ i , …, σ r ): is the diagonal matrix of the eigenvalues ​​of X. Applying the proximal mapping of formula (3) to the quadratic partial gradient of formula (2) yields: Ψ1 and Ψ2 are dictionary matrices, and the minimum value of formula (1) is obtained: Or it can be expressed as formula (4): L f is the Lipschitz constant of the quadratic term in formula (2), that is, A H The spectral norm of A, the kth iteration is used as the kth layer of the feedforward network in the through-wall radar imaging model, and the through-wall radar imaging model is formed according to formula (4), replacing each matrix that depends on Ψ1 and Ψ2 with a convolutional layer of corresponding size The equation for the kth layer is: Wherein, * represents the corresponding convolution operator in the through-wall radar imaging model.

2. The through-wall radar imaging method according to claim 1, characterized in that: The obtaining of radar echo data in a preset radar detection scene includes: A single-station stepped frequency system is adopted, and the co-located transmitting and receiving antenna is stepped forward from left to right in a direction parallel to the wall, and a first preset number of measuring positions are set in the direction of the measuring line, and each of the measuring positions transmits a second preset number of frequency point signals; The radar echo data measured at the nth frequency signal and the mth measurement position is Z n,m ; is the wall reflection signal in the preset radar detection scene, is the target reflection signal in the preset radar detection scene, v n，m A measurement noise signal in the preset radar detection scene; The wall reflects the signal for: The target reflected signal for: Among them, σ w is the reflection coefficient of the wall in the preset radar detection scene, f n is the mth frequency signal, τ w represents the round-trip transmission delay from the mth measurement position to the wall, which is independent of the measurement position, p is the total number of targets, σ p is the reflection coefficient of the pth target, τ m，q represents the round-trip delay between the mth measurement location and the qth target; The total echo signal in the radar echo data is: Z=Z w +Z t +V.

3. The through-wall radar imaging method according to claim 1, characterized in that: The discretizing the imaging scene in the preset radar detection scene includes: The imaging scene is two-dimensionally discretized to obtain discrete grids, and the total number of discrete grids Q=KxL′, wherein K and L′ are the numbers of discrete grids corresponding to the range direction and the azimuth direction in the imaging scene, respectively.

4. The through-wall radar imaging method according to claim 3, characterized in that: The step of constructing a dictionary matrix according to the discretization result of the imaging scene comprises: Performing a column stacking operation on the discrete grids to obtain a scene reflectivity vector; The qth element S in the scene reflectivity vector q for: The target reflection signal is: Z t =Ψ1S Ψ1 is a dictionary matrix of MN×Q dimensions: Ψ1=exp(-j2πf n t m，p ) q=1,2,...,Q,f n is the nth frequency signal, p is the total number of targets, τ m，q represents the round-trip delay between the mth measurement location and the qth target; S is a sparse vector, the qth element of the reflectivity vector L of the wall q It is expressed as: The wall reflection signal is: WITH w =Ψ2L Ψ2 is a dictionary matrix of MN×Q dimensions: Ψ2=exp(-j2πf n t w ) τ w represents the round-trip transmission delay from the mth measurement location to the wall, and L is a low-rank matrix with wall reflection.

5. The through-wall radar imaging method according to claim 4, characterized in that: The low-rank sparse signal model constructed according to the dictionary matrix is: D=Ψ1L+Ψ2S+V L is a low-rank matrix with wall reflections, S is a sparse matrix with target reflections, N is a measurement noise matrix, and Ψ1 and Ψ2 are the dictionary matrices.

6. A through-the-wall radar imaging system for implementing the method according to any one of claims 1 to 5, characterized in that: The through-wall radar imaging system comprises: A radar echo data acquisition unit, used to acquire radar echo data in a preset radar detection scene, and discretize an imaging scene in the preset radar detection scene; A signal model construction unit, used to construct a dictionary matrix according to the discretization result of the imaging scene, and to construct a low-rank sparse signal model according to the dictionary matrix, wherein the low-rank sparse signal model is used to characterize the low-rank characteristics of the wall clutter and the spatial sparse characteristics of the target behind the wall to be detected in the preset radar detection scene; A low-rank sparse problem determination unit, used to determine a low-rank sparse optimization problem in the low-rank sparse signal model, and construct a through-wall radar imaging model according to the determined low-rank sparse optimization problem; The imaging analysis unit is used to perform model training on the through-wall radar imaging model until the through-wall radar imaging model converges, and input the echo data to be imaged into the converged through-wall radar imaging model for imaging analysis to obtain a reconstructed image.

7. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images