Interference three-dimensional imaging method for inverse synthetic aperture radar

Abstract

The invention is suitable for the technical field of radar signal processing, and provides an inverse synthetic aperture radar interference three-dimensional imaging method, which comprises the following steps: receiving an echo signal corresponding to a linear frequency modulation signal transmitted by a radar, preprocessing the echo signal to obtain multi-channel echo data, and then according to the multi-channel echo data, calculating a linear frequency modulation signal corresponding to the linear frequency modulation signal; the method comprises the steps of constructing a structured low-rank matrix, reweighting image domain data of multi-channel echo data by using an L21 norm, constructing a multi-channel structured low-rank and sparse model, decomposing the model through a Hankel matrix to obtain a first initial value and a second initial value, and solving the multi-channel structured low-rank and sparse model based on the first initial value, the second initial value and the L21 norm. And finally, obtaining an inverse synthetic aperture radar interference three-dimensional image based on the multi-channel image. Therefore, when the distance Doppler method is used for obtaining the solution of the image, the calculation complexity is greatly reduced, and the imaging efficiency is improved.

Description

Technical Field

[0001] The present application belongs to the field of radar signal processing technology, and in particular relates to a method for inverse synthetic aperture radar interferometry three-dimensional imaging. Background Art

[0002] Interferometric inverse synthetic aperture radar (InISAR) is an advanced multi-channel sensor that can perform three-dimensional imaging of moving targets such as aircraft and satellites. It has been widely used in remote sensing and other fields.

[0003] The success of InISAR imaging depends largely on the fact that imaging processing must maintain high coherence between multi-channel data to achieve high-quality interferometric phase recovery. In modern advanced radar scenarios, compressed data sampling of sparse frequency bands (SFBs) and sparse apertures (SAs) may be encountered. In practical applications, due to the loss of trajectory tracking and discontinuous observations of multiple targets by multi-function radars, SA measurements are unavoidable for ISAR imaging. In addition, the transmission of sparse frequency band signals, namely SFBs, has been used for intelligent anti-interference in complex electromagnetic environments. The presence of SFBs and SAs often significantly increases the difficulty of InISAR imaging. Summary of the Invention

[0004] The embodiments of the present application provide a method for inverse synthetic aperture radar interferometry three-dimensional imaging, which can solve the current problem of difficult InISAR imaging caused by sparse frequency bands and sparse apertures.

[0005] In a first aspect, an embodiment of the present application provides a method for inverse synthetic aperture radar interference three-dimensional imaging, comprising: receiving an echo signal corresponding to a linear frequency modulation signal emitted by a radar, and preprocessing the echo signal to obtain multi-channel echo data; constructing a structured low-rank matrix based on the multi-channel echo data, reweighting the image domain data of the multi-channel echo data using the L21 norm, and constructing a multi-channel structured low-rank and sparse model, wherein the structured low-rank matrix is ​​a two-layer Hankel matrix; decomposing the model through a Hankel matrix to obtain a first initial value and a second initial value; solving the multi-channel structured low-rank and sparse model based on the first initial value, the second initial value and the L21 norm to obtain a multi-channel image suitable for interference processing; and obtaining an inverse synthetic aperture radar interference three-dimensional image based on the multi-channel image.

[0006] In a possible implementation of the first aspect, the receiving an echo signal corresponding to the linear frequency modulation signal transmitted by the radar and preprocessing the echo signal to obtain multi-channel echo data includes:

[0007] Performing linear frequency modulation processing on the echo signal to obtain a demodulated echo signal;

[0008] Perform range compression and azimuth Fourier transform on the demodulated echo signal to obtain the frequency domain echo signal:

[0009]

[0010] in, λ represents wavelength, γ represents modulation frequency, T a and f d denote the coherent processing interval (CPI) and Doppler frequency domain, T denotes the pulse width, and f r represents the distance frequency domain, c represents the speed of light, i∈{O,A,B} represents different channels, i.e. different receiving radars, σ i,p represents the scattering coefficient of the pth scatterer in the i-th channel, x p and z p represents the position of the scatterer, ω x and ω z Represents the rotation speed around the X axis and Z axis respectively, R p Indicates the distance from the scatterer to the radar;

[0011] Perform sparse sampling on the frequency domain echo signal to obtain multi-channel echo data:

[0012]

[0013] Among them, j represents the imaginary unit, π represents the circumference of a circle, and F s represents the sampling frequency, PRF represents the pulse repetition frequency, m=1,2,...,M represents the fast time sequence in discrete form, n=1,2,...,N represents the slow time sequence in discrete form, and S(m,n) represents the two-dimensional echo data after sparse sampling.

[0014] Optionally, in another possible implementation of the first aspect, constructing a structured low-rank matrix based on multi-channel echo data, reweighting the image domain data using the L21 norm, and constructing a multi-channel structured low-rank and sparse model include:

[0015] Construct the Hankel matrix based on the nth column Y(:,n) of the multi-channel echo data:

[0016]

[0017] Among them, Y i Indicates all data to be recovered for the i-th channel;

[0018] Construct a two-layer Hankel matrix using all columns of multi-channel echo data:

[0019]

[0020] Where P and Q are beam parameters;

[0021] According to the image domain data, sparse constraints based on L21 norm are performed:

[0022]

[0023] F 2D (X i )=Y i ,i∈{O,A,B}

[0024] Among them, ||·|| 2,1 represents the L21 norm of the matrix, that is W⊙X is the result of L21 norm reweighting, W represents the weighted vector, each element of which is represented by That is, the inverse of the corresponding element of x, usually X i represents the ISAR image domain data of the i-th channel, F 2D (·) represents the Fourier transform of the distance and azimuth dimensions, and ⊙ is the Hadamard product;

[0025] Using the weight parameter, the proportion of balanced sparse constraints and low-rank constraints is expressed as:

[0026]

[0027] F 2D (X i )=Y i ,i∈{O,A,B}

[0028]

[0029] Among them, ||·|| * represents the nuclear norm of the matrix, S i represents the matrix corresponding to the echo data under sparse sampling of the i-th channel, Ω represents the sampling rate, κ represents the constant coefficient, which is used to balance the sparse constraint and the low-rank constraint, and 0<κ<1.

[0030] Optionally, in another possible implementation of the first aspect, decomposing the model through a Hankel matrix to obtain the first initial value and the second initial value includes:

[0031] Decompose the two-layer Hankel matrix into the product of two parts, that is, H{Y}=UV H ;

[0032] Use the low-rank matrix fitting method to estimate the rank and get the first initial value U (0) and the second initial value V(0) ;

[0033] Among them, U, V H represents the matrix after the second-level Hankel matrix decomposition.

[0034] Optionally, in another possible implementation of the first aspect, solving the multi-channel structured low-rank and sparse model based on the first initial value, the second initial value, and the L21 norm to obtain a multi-channel image suitable for interference processing includes:

[0035] Based on the first and second initial values, construct an optimization problem with constraints:

[0036]

[0037] stH(Y)=UV H

[0038] F 2D (X i )=Y i ,i∈{O,A,B}

[0039]

[0040] Among them, ||·|| F represents the F norm;

[0041] The alternating iterative multiplier method based on the augmented Lagrangian function is used to solve the optimization problem with constraints to fill the structured low-rank matrix and obtain the filled structured low-rank matrix, where the augmented Lagrangian function is expressed as:

[0042]

[0043] Among them, R and R i Denote the dual variables, ρ and ρ i represents the penalty coefficient;

[0044] Perform an inverse transformation on the padded structured low-rank matrix to obtain an inverse transformation matrix;

[0045] The inverse transformation matrix is ​​imaged using the range-Doppler method to obtain a multi-channel image suitable for interferometric processing.

[0046] Optionally, in another possible implementation manner of the first aspect, the iterative steps of the alternating iterative multiplier method are expressed as:

[0047] U (k+1) =ρ(H(Y (k) )+R (k) )·V (k) ·(κI+ρV (k)HV (k) ) -1

[0048] V (k+1) =ρ(H(Y (k) )+R (k) )·U (k+1) ·(κI+ρU (k+1)H U (k+1) ) -1

[0049]

[0050] R (k+1) =H(Y (k+1) )-U (k+1) V (k+1)H +R (k)

[0051] R i (k+1) =Y i (k+1) -F 2D (X i (k+1) )+R i (k)

[0052] in,(·) k and(·) k+1 denote the variables obtained in the kth and k+1th iterations, respectively, I denotes the identity matrix, represents the inverse structural operation, P[(·) i ]express[(·) O (·) A (·) B ].

[0053] Optionally, in another possible implementation of the first aspect, obtaining an inverse synthetic aperture radar interferometry three-dimensional image based on a multi-channel image includes:

[0054] Based on the multi-channel image, scatterers with scattering greater than a preset threshold are selected;

[0055] Estimation of rotation parameters based on scatterers;

[0056] The threshold parameters are adjusted to perform outlier detection and clutter suppression to obtain the inverse synthetic aperture radar interferometry 3D image.

[0057] Optionally, in another possible implementation of the first aspect, the coordinate calculation of the scatterer is expressed as:

[0058]

[0059] Among them, n p represents the scatterer of the pth distance unit, R A,p and R B,p are the distances from the scatterer to radar A and radar B respectively, and represents the interference phase difference, L A Indicates the baseline length.

[0060] Optionally, in another possible implementation manner of the first aspect, the rotation parameter estimation based on the scatterer is expressed as:

[0061]

[0062] Among them, W T is a digital weighting matrix that can be obtained by the amplitude of each scatterer, f=[f1 f2...f P ] T represents the Doppler frequency of each scatterer,

[0063] Optionally, in another possible implementation of the first aspect, the threshold parameter is expressed as:

[0064]

[0065] Among them, δ d is the threshold parameter, which is related to the clutter and noise level of the selected sample.

[0066] The present application provides a method for inverse synthetic aperture radar interferometry three-dimensional imaging. The method first receives an echo signal corresponding to a linear frequency modulation signal emitted by a radar and preprocesses the echo signal to obtain multi-channel echo data. A structured low-rank matrix is ​​then constructed based on the multi-channel echo data. The image domain data of the multi-channel echo data is reweighted using the L21 norm to construct a multi-channel structured low-rank and sparse model. The model is then decomposed through a Hankel matrix to obtain first and second initial values. The multi-channel structured low-rank and sparse model is solved based on the first and second initial values ​​and the L21 norm to obtain a multi-channel image suitable for interference processing. Finally, an inverse synthetic aperture radar interferometry three-dimensional image is obtained based on the multi-channel image. As a result, the computational complexity of solving the image obtained using the range Doppler method is greatly reduced, and the imaging efficiency is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0068] Figure 1 A flowchart of a method for inverse synthetic aperture radar interferometry 3D imaging provided for an embodiment of the present application;

[0069] Figure 2(a) to Figure 2(d) A comparison chart of imaging results using different sparse modes provided in the embodiments of the present application and other typical imaging methods;

[0070] Figure 3(a) to Figure 3(b) This is a comparison chart of the 3D reconstruction results under different sparse modes provided in the embodiments of the present application and the 3D reconstruction results of other typical methods. DETAILED DESCRIPTION

[0071] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0072] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0073] It will also be understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0074] As used in this specification and the appended claims, the term "if" can be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0075] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0076] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present application. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more but not all embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0077] The following is a description of the technical terms involved in this application:

[0078] Low-rank matrix recovery (LRMR) is widely used in image processing for image restoration, such as denoising and deblurring. The data matrix of a clear natural image is often low-rank or nearly low-rank, but the presence of random errors of arbitrarily large magnitude but sparse distribution destroys the low-rank nature of the original data. Low-rank matrix recovery treats the degraded image as a set of low-dimensional data with noise added, so that the original data can be approximated by a low-rank matrix.

[0079] Low rank refers to a matrix with a small rank, while sparse refers to a matrix with a small number of non-zero elements. If a matrix is ​​subjected to singular value decomposition and all its singular values ​​are arranged into a vector, the sparsity of this vector corresponds to the low rank of the matrix.

[0080] Rank, suppose there is an r-order minor D in the matrix A that is not equal to 0, and all r+1-order minors (if any) are equal to 0, then D is called the highest-order non-zero minor of the matrix A, and the number r is called the rank of the matrix A, denoted by R(A), and it is stipulated that the rank of the zero matrix is ​​equal to 0.

[0081] L21 norm-based reweighting: It is an iterative optimization method that enhances the sparsity of matrix rows by continuously adjusting weights to enable more efficient feature selection and sparse signal recovery.

[0082] Compressed sensing (CS) methods have been widely used to address existing issues with sparse imaging. Targets are typically assumed to consist of a finite number of scattering centers, due to their sparse nature. Leveraging this prior knowledge of sparsity, CS methods have demonstrated promising performance in sparse imaging applications. However, these methods inevitably require the construction of discrete dictionaries, typically Fourier transform dictionaries. The model mismatch introduced by gridding is a major drawback of CS methods.

[0083] For example, a gridless sparse imaging method is matrix filling. This method uses low-rank properties to accurately restore a matrix even if some elements are missing, under certain conditions. Matrix filling requires not only that the echo data possess low-rank properties but also that it satisfy certain incoherence criteria. Low-rank properties are also limited in some practical applications because they rely heavily on random missing samples (RMS) data sampling and struggle to work with gap missing samples (GMS).

[0084] Therefore, to address the above issues, this application provides a method for inverse synthetic aperture radar interferometry 3D imaging. This method proposes a joint sparse and low-rank matrix reconstruction method, constructs a new multi-channel Hankel matrix model, and uses non-convex weighted sparse constraints to improve performance. Joint sparsity enhancement is applied to multi-channel images, rather than independent channel processing, to maintain high coherence. The method utilizes an alternating iterative multiplier method for solving the problem, effectively improving sparse imaging performance. A post-processing optimization method for the sparsely reconstructed images is then used to estimate 3D target geometry.

[0085] The following describes in detail a method for inverse synthetic aperture radar interferometry three-dimensional imaging provided by the present application with reference to the accompanying drawings.

[0086] Figure 1 A flow chart of a method for inverse synthetic aperture radar interferometry three-dimensional imaging provided in an embodiment of the present application is shown.

[0087] Step 101: receiving an echo signal corresponding to a linear frequency modulation signal transmitted by a radar, and preprocessing the echo signal to obtain multi-channel echo data;

[0088] Furthermore, in the embodiment of the present application, the above step 101 includes:

[0089] Performing linear frequency modulation processing on the echo signal to obtain a demodulated echo signal;

[0090] Perform range compression and azimuth Fourier transform on the demodulated echo signal to obtain the frequency domain echo signal:

[0091]

[0092] in, λ represents wavelength, γ represents modulation frequency, T a and f d denote the coherent processing interval (CPI) and Doppler frequency domain, T denotes the pulse width, and f r represents the distance frequency domain, c represents the speed of light, i∈{O , A, B} represent different channels, i.e. different receiving radars, σ i,p represents the scattering coefficient of the pth scatterer in the i-th channel, x p and z p represents the position of the scatterer, ω x and ω z Represents the rotation speed around the X axis and Z axis respectively, R p Indicates the distance from the scatterer to the radar;

[0093] Perform sparse sampling on the frequency domain echo signal to obtain multi-channel echo data:

[0094]

[0095] Among them, j represents the imaginary unit, π represents the circumference of a circle, and F s represents the sampling frequency, PRF represents the pulse repetition frequency, m=1,2,...,M represents the fast time sequence in discrete form, n=1,2,...,N represents the slow time sequence in discrete form, and S(m,n) represents the two-dimensional echo data after sparse sampling.

[0096] Step 102: constructing a structured low-rank matrix based on the multi-channel echo data, reweighting the image domain data of the multi-channel echo data using the L21 norm, and constructing a multi-channel structured low-rank and sparse model, wherein the structured low-rank matrix is ​​a two-layer Hankel matrix;

[0097] Furthermore, in the embodiment of the present application, the above step 102 includes:

[0098] Construct the Hankel matrix based on the nth column Y(:,n) of the multi-channel echo data:

[0099]

[0100] Among them, Y i Indicates all data to be recovered for the i-th channel;

[0101] Construct a two-layer Hankel matrix using all columns of multi-channel echo data:

[0102]

[0103] Where P and Q are beam parameters;

[0104] According to the image domain data, sparse constraints based on L21 norm are performed:

[0105]

[0106] Among them, ||·|| 2,1 represents the L21 norm of the matrix, that is W⊙X is the result of L21 norm reweighting, W represents the weighted vector, each element of which is represented by That is, the inverse of the corresponding element of x, usually X i represents the ISAR image domain data of the i-th channel, F 2D (·) represents the Fourier transform of the distance and azimuth dimensions, and ⊙ is the Hadamard product;

[0107] Using the weight parameter, the proportion of balanced sparse constraints and low-rank constraints is expressed as:

[0108]

[0109] Among them, ||·|| * represents the nuclear norm of the matrix, S i represents the matrix corresponding to the echo data under sparse sampling of the i-th channel, Ω represents the sampling rate, κ represents the constant coefficient, which is used to balance the sparse constraint and the low-rank constraint, and 0<κ<1.

[0110] It is understood that the low-rank property of the Hankel matrix can be exploited for sparse ISAR imaging of targets consisting of a limited number of scattering points. In the matrix bundle method, the Hankel matrix is ​​constructed to fully exploit the signal's translational invariance for high-resolution spectral estimation while minimizing the effects of noise. Therefore, the two-layer Hankel structured method constructed in this application can enhance the low-rank prior, and the derived results are more conducive to sparse ISAR imaging.

[0111] Step 103, decomposing the model through Hankel matrix to obtain a first initial value and a second initial value;

[0112] Furthermore, in the embodiment of the present application, the above step 103 includes:

[0113] Decompose the two-layer Hankel matrix into the product of two parts, that is, H{Y}=UV H ;

[0114] Use the low-rank matrix fitting method to estimate the rank and get the first initial value U (0) and the second initial value V (0) ;

[0115] Among them, U, V H represents the matrix after the second-level Hankel matrix decomposition.

[0116] Step 104: solving a multi-channel structured low-rank and sparse model based on the first initial value, the second initial value and the L21 norm to obtain a multi-channel image suitable for interference processing;

[0117] Furthermore, in the embodiment of the present application, the above step 104 includes:

[0118] Step 1041: Based on the first initial value and the second initial value, construct an optimization problem with constraints:

[0119]

[0120] Among them, ||·|| F represents the F norm;

[0121] Step 1042: Solve the constrained optimization problem using an alternating iterative multiplier method based on an augmented Lagrangian function to fill the structured low-rank matrix and obtain a filled structured low-rank matrix. The augmented Lagrangian function is expressed as:

[0122]

[0123] Among them, R and R i Denote the dual variables, ρ and ρ i represents the penalty coefficient;

[0124] Among them, the iterative steps of the alternating iterative multiplier method are expressed as:

[0125] U (k+1) =ρ(H(Y (k) )+R (k) )·V (k) ·(κI+ρV (k)H V (k) ) -1 (9)

[0126] V (k+1) =ρ(H(Y (k) )+R (k) )·U (k+1) ·(αI+ρU (k+1)H U (k+1) ) -1 (10)

[0127]

[0128] R (k+1) =H(Y (k+1) )-U (k+1) V (k+1)H +R (k) (13)

[0129] R i (k+1) =Y i (k+1) -F 2D (X i (k+1) )+R i (k) (14)

[0130] in,(·) k and(·) k+1 denote the variables obtained in the kth and k+1th iterations, respectively, I denotes the identity matrix, represents the inverse structural operation, P[(·) i ]express[(·) O (·) A (·) B ].

[0131] Preferably, in one embodiment, each iteration of the alternating iterative multiplier method is to sequentially perform the multiplication of U, V, Y, X, R, R i Find the point where the conjugate gradient is 0. Because this function is about U, V, Y, X, R, R i All three variables are convex, so the extreme point is the maximum point. The iterative steps of the above alternating iterative multiplier method are as follows:

[0132] Combining the above formula (9) with formula (8), we get formula (15):

[0133]

[0134] The gradient of equation (15) is calculated with respect to the conjugate U, namely:

[0135]

[0136] Based on the same method as above, the above equations (10) and (11) can be solved.

[0137] For the above formula (12), combined with formula (8), we can get the reweighted L21 norm, namely:

[0138]

[0139] Among them, x ti =X(t,i). The Frobenius norm of each channel is expressed as:

[0140]

[0141] Among them, F r F dand C represent the range and azimuth FFT matrices and the items that are independent of X, respectively.

[0142] Combining equations (17) and (18), we get equation (19):

[0143]

[0144] in, P[(·) i ]=[(·) O (·) A (·) B ], i∈{O,A,B}, Z represents a diagonal matrix.

[0145] Finally, the update of auxiliary variables R and Ri is a fixed method under the framework of alternating iterative multiplier method, namely, equations (13) and (14):

[0146] R (k+1) =H(Y (k+1) )-U (k+1) V (k+1)H +R (k) (13)

[0147] R i (k+1) =Y i (k+1) -F 2D (X i (k+1) )+R i (k) (14)

[0148] In the above description, since there is a fixed formula for calculating the conjugate gradient of a matrix, the specific calculation steps will not be described in detail in this application.

[0149] Step 1043, performing an inverse transformation on the padded structured low-rank matrix to obtain an inverse transformation matrix;

[0150] Specifically, the structured matrix obtained by solving Perform inverse transformation to obtain the inverse transformation matrix X. The inverse transformation is to transform the structured Hankel matrix back to the original two-dimensional echo matrix, and its elements are one-to-one corresponding.

[0151] Step 1044 , performing imaging processing on the inverse transformation matrix using the range-Doppler method to obtain a multi-channel image suitable for interference processing.

[0152] Optionally, the range-Doppler method performs ISAR imaging processing on the inverse transformation matrix X in a dechirp mode, which is actually a two-dimensional Fourier transform, namely, a 2D-FFT.

[0153] Step 105: Obtain an inverse synthetic aperture radar interferometry three-dimensional image based on the multi-channel image.

[0154] Furthermore, in the embodiment of the present application, the above step 105 includes:

[0155] Step 1051 , selecting a scatterer having a scattering property greater than a preset threshold based on the multi-channel image;

[0156] In the embodiment of the present application, the scatterer coordinate calculation is expressed as:

[0157]

[0158] Among them, n p represents the scatterer of the pth distance unit, R A,p and R B,p are the distances from the scatterer to radar A and radar B respectively, and represents the interference phase difference, L A Indicates the baseline length.

[0159] Step 1052, estimating rotation parameters based on the scatterer;

[0160] In the embodiment of the present application, the rotation parameter is estimated based on the scatterer, which is expressed as:

[0161]

[0162] Among them, W T is a digital weighting matrix that can be obtained by the amplitude of each scatterer, f=[f1 f2...f P ] T represents the Doppler frequency of each scatterer,

[0163] Step 1053 , adjusting the threshold parameters, performing outlier detection and clutter suppression, and obtaining an inverse synthetic aperture radar interferometry 3D image.

[0164] In the embodiment of the present application, the threshold parameter is expressed as:

[0165]

[0166] Among them, δ d is the threshold parameter, which is related to the clutter and noise level of the selected sample.

[0167] The present application provides a method for inverse synthetic aperture radar interferometry three-dimensional imaging. The method first receives an echo signal corresponding to a linear frequency modulation signal emitted by a radar and preprocesses the echo signal to obtain multi-channel echo data. A structured low-rank matrix is ​​then constructed based on the multi-channel echo data. The image domain data of the multi-channel echo data is reweighted using the L21 norm to construct a multi-channel structured low-rank and sparse model. The model is then decomposed through a Hankel matrix to obtain first and second initial values. The multi-channel structured low-rank and sparse model is solved based on the first and second initial values ​​and the L21 norm to obtain a multi-channel image suitable for interference processing. Finally, an inverse synthetic aperture radar interferometry three-dimensional image is obtained based on the multi-channel image. As a result, the computational complexity of solving the image obtained using the range Doppler method is greatly reduced, and the imaging efficiency is improved.

[0168] To illustrate the effectiveness of this application for inverse synthetic aperture radar interferometry 3D imaging under various sparse sampling modes, experiments based on measured data are further demonstrated:

[0169] (1) Experimental setup

[0170] The measured data is based on a civil aircraft. The system has a center frequency of 8.1 GHz, a bandwidth of 800 MHz for transmitting linear frequency modulation signals, a pulse repetition frequency of 150 Hz, and operates in a delinear frequency modulation mode, using 256 pulses for inverse synthetic aperture imaging.

[0171] In order to verify the effectiveness of the present invention, different sparse sampling methods are adopted, the sparse rate is set to 0.4, and the imaging quality is compared.

[0172] (2) Experimental content

[0173] Based on the MATLAB software platform, different sampling modes were adopted to compare the imaging results of the present invention with those of the structured low-rank matrix completion method and the compressed sensing method based on the L1 norm.

[0174] Figure 2(a) to Figure 2(d) Figure 2(a) shows the sparse imaging results of different sparse sampling modes, namely, sparse dictionaries; Figure 2(b) shows the imaging results under the two-dimensional random sampling mode; Figure 2(c) shows the imaging results under the two-dimensional block sampling mode; Figure 2(d) shows the imaging results under the combined two-dimensional random sampling and two-dimensional block sampling modes.

[0175] Figure 2 shows the sparse imaging results under different sparse sampling modes, from top to bottom: the imaging effects of the compressed sensing method based on the L1 norm, the structured low-rank prior algorithm, and the ISAR imaging algorithm of this application.

[0176] Figure 3 (a) and (b) are the three-dimensional target reconstruction results with a sparse sampling rate of 0.4 under the two-dimensional random sampling mode and the joint two-dimensional random sampling and two-dimensional block sampling mode. From top to bottom are the three-dimensional target reconstruction effects of the compressed sensing method based on the L1 norm, the structured low-rank prior algorithm and the ISAR imaging algorithm of this application. Among them, the multi-channel structured low-rank sparse method proposed in this application has a more distinguishable three-dimensional reconstructed shape than other images. This conclusion is consistent with Figure 2(a) to Figure 2(d) The intuitive image quality is consistent.

[0177] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0178] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for inverse synthetic aperture radar interferometry three-dimensional imaging, characterized in that: include: receiving an echo signal corresponding to a linear frequency modulation signal transmitted by a radar, and preprocessing the echo signal to obtain multi-channel echo data; Constructing a structured low-rank matrix based on the multi-channel echo data, reweighting the image domain data of the multi-channel echo data using the L21 norm, and constructing a multi-channel structured low-rank and sparse model, wherein the structured low-rank matrix is ​​a two-layer Hankel matrix; Decomposing the model through Hankel matrix to obtain a first initial value and a second initial value; Solving the multi-channel structured low-rank and sparse model based on the first initial value, the second initial value and the L21 norm to obtain a multi-channel image suitable for interference processing; An inverse synthetic aperture radar interference three-dimensional image is obtained based on the multi-channel image.

2. The inverse synthetic aperture radar interferometry 3D imaging method according to claim 1, wherein: The receiving radar transmits an echo signal corresponding to a linear frequency modulation signal and preprocessing the echo signal to obtain multi-channel echo data, including: performing a delinear frequency modulation process on the echo signal to obtain a demodulated echo signal; The demodulated echo signal is subjected to range compression and azimuth-dimensional Fourier transform to obtain a frequency domain echo signal: in, λ represents wavelength, γ represents modulation frequency, T a and f d denote the coherent processing interval (CPI) and Doppler frequency domain, T denotes the pulse width, and f r represents the distance frequency domain, c represents the speed of light, i∈{O,A,B} represents different channels, i.e. different receiving radars, σ i,p represents the scattering coefficient of the pth scatterer in the i-th channel, x p and z p represents the position of the scatterer, ω x and ω z Represents the rotation speed around the X axis and Z axis respectively, R p Indicates the distance from the scatterer to the radar; Perform sparse sampling processing on the frequency domain echo signal to obtain the multi-channel echo data: Among them, j represents the imaginary unit, π represents the circumference of a circle, and F s represents the sampling frequency, PRF represents the pulse repetition frequency, m=1,2,...,M represents the fast time sequence in discrete form, n=1,2,...,N represents the slow time sequence in discrete form, and S(m,n) represents the two-dimensional echo data after sparse sampling.

3. The inverse synthetic aperture radar interferometry 3D imaging method according to claim 2, wherein: The method comprises: constructing a structured low-rank matrix based on the multi-channel echo data, reweighting the image domain data using the L21 norm, and constructing a multi-channel structured low-rank and sparse model. The Hankel matrix is ​​constructed according to the nth column Y(:,n) of the multi-channel echo data: Among them, Y i Indicates all data to be recovered for the i-th channel; A two-layer Hankel matrix is ​​constructed using all columns of the multi-channel echo data: Where P and Q are beam parameters; According to the image domain data, a sparse constraint based on the L21 norm is performed: s.t.F 2D (X i )=Y i ,i∈{O,A,B} Among them, ||·|| 2,1 represents the L21 norm of the matrix, that is W⊙X is the result of L21 norm reweighting, W represents the weighted vector, each element of which is represented by That is, the inverse of the corresponding element of x, usually ò=10 -5 , X i represents the ISAR image domain data of the i-th channel, F 2D (·) represents the Fourier transform of the distance and azimuth dimensions, and ⊙ is the Hadamard product; Using the weight parameter, the proportion of balanced sparse constraints and low-rank constraints is expressed as: s.t.F 2D (X i )=Y i ,i∈{O,A,B} Among them, ||·|| * represents the nuclear norm of the matrix, S i represents the matrix corresponding to the echo data under sparse sampling of the i-th channel, Ω represents the sampling rate, κ represents the constant coefficient, which is used to balance the sparse constraint and the low-rank constraint, and 0<κ<1.

4. The inverse synthetic aperture radar interferometry 3D imaging method according to claim 3, wherein: Decomposing the model through a Hankel matrix to obtain a first initial value and a second initial value includes: Decompose the two-layer Hankel matrix into the product of two parts, namely H{Y}=UV H ; Use the low-rank matrix fitting method to estimate the rank and get the first initial value U (0) and the second initial value V (0) ; Among them, U, V H represents the matrix after the two-layer Hankel matrix decomposition.

5. The inverse synthetic aperture radar interferometry 3D imaging method according to claim 4, wherein: Solving the multi-channel structured low-rank and sparse model based on the first initial value, the second initial value and the L21 norm to obtain a multi-channel image suitable for interference processing includes: Based on the first initial value and the second initial value, a constrained optimization problem is constructed: s.t.H(Y)=UV H F 2D (X i )=Y i ,i∈{O,A,B} Among them, ||·|| F represents the F norm; The constrained optimization problem is solved by using an alternating iterative multiplier method based on an augmented Lagrangian function to fill the structured low-rank matrix to obtain a filled structured low-rank matrix, wherein the augmented Lagrangian function is expressed as: Among them, R and R i Denote the dual variables, ρ and ρ i represents the penalty coefficient; Performing an inverse transformation on the padded structured low-rank matrix to obtain an inverse transformation matrix; The inverse transformation matrix is ​​imaged using a range-Doppler method to obtain the multi-channel image suitable for interference processing.

6. The inverse synthetic aperture radar interferometry 3D imaging method according to claim 5, characterized in that: The iterative steps of the alternating iterative multiplier method are expressed as: in,(·) k and(·) k+1 denote the variables obtained in the kth and k+1th iterations, respectively, I denotes the identity matrix, represents the inverse structural operation, P[(·) i ]express[(·) O (·) A (·) B ].

7. The inverse synthetic aperture radar interferometry 3D imaging method according to any one of claims 1 to 6, characterized in that: The step of obtaining an inverse synthetic aperture radar interferometry three-dimensional image based on the multi-channel image includes: Based on the multi-channel image, selecting a scatterer having a scattering property greater than a preset threshold; Estimating rotation parameters based on the scatterer; The threshold parameters are adjusted to perform outlier detection and clutter suppression to obtain the inverse synthetic aperture radar interference three-dimensional image.

8. The inverse synthetic aperture radar interferometry 3D imaging method according to claim 7, characterized in that: The coordinate calculation of the scatterer is expressed as: Among them, n p represents the scatterer of the pth distance unit, R A,p and R B,p are the distances from the scatterer to radar A and radar B respectively, and represents the interference phase difference, L A Indicates the baseline length.

9. The inverse synthetic aperture radar interferometry 3D imaging method according to claim 8, characterized in that: The rotation parameter estimation is performed based on the scatterer, which is expressed as: Among them, W T is a digital weighting matrix that can be obtained by the amplitude of each scatterer, f=[f1 f2...f P ] T represents the Doppler frequency of each scatterer, 10. The inverse synthetic aperture radar interferometry 3D imaging method according to claim 9, characterized in that: The threshold parameter is expressed as: Among them, δ d is the threshold parameter, which is related to the clutter and noise level of the selected sample.

Images