CN107167805A - Based on the common sparse ISAR high-resolution imaging method of multilayer - Google Patents

Abstract

A high-resolution imaging method based on multi-layer co-sparse inverse synthetic aperture radar, the imaging process is: input the echo data after motion compensation and range compression; learn the analytical operator of each range unit of the echo signal , and then denoise the echo data; after obtaining the optimal analytical operator and the denoised echo signal, the present invention adopts the modified OMP algorithm to restore the ISAR image; in the whole imaging process, the imaging process of the present invention is divided into For multiple stages, the low-resolution ISAR image obtained in the previous stage is transformed into the data domain through inverse Fourier transform and used as the input of the next stage. The invention can ensure that the azimuth resolution of each stage is improved compared with the previous stage. The invention is used for ISAR target imaging in sparse scenes and has strong robustness to noise.

Description

Inverse synthetic aperture radar high-resolution imaging method based on multi-layer co-sparse

technical field

The invention belongs to the technical field of communication, and further relates to a multi-layer co-sparse based inverse synthetic aperture radar high-resolution imaging method in the technical field of radar signal processing. The present invention is aimed at the high-resolution imaging method of sparse ISAR targets within a short CPI, and adopts a multi-layer co-sparse analytical model to realize high-resolution inverse synthetic aperture radar imaging, so as to solve the problem that the traditional ISAR imaging method is not robust to noise, and can be applied In the later target classification and recognition work.

Background technique

In actual military applications, high-resolution inverse synthetic aperture radar ISAR images can provide great convenience for subsequent target recognition and classification work, so high-resolution inverse synthetic aperture radar ISAR images have great practical significance. The resolution of a radar image consists of two azimuth resolutions: range resolution and azimuth resolution. The bandwidth of the transmitted signal of the radar system determines the range resolution. As long as the radar system is fixed, the range resolution is fixed. In general, the wider the transmit signal bandwidth, the better the range resolution. Azimuth resolution is related to the coherent accumulation time of radar and target. Generally speaking, the longer the coherent accumulation time is, the more target information can be obtained, so the azimuth resolution will be better. Generally, if you want to obtain high azimuth resolution, you need a long coherent integration time (CPI). However, due to the strong maneuverability and non-cooperation of the inverse synthetic aperture radar ISAR target, its motion state is very complicated, which brings great difficulties to motion compensation, and it is difficult for the radar system to achieve accurate tracking of the target for a long time. Therefore, it is of great practical significance to realize high-resolution imaging of inverse synthetic aperture radar (ISAR) in a short observation time.

Liu, H.C., Jiu, B., Liu, H.W. and Bao, Z. Four authors published the paper "Superresolution ISAR imaging based on sparse Bayesian learning." (IEEETransactions on Geoscience and Remote Sensing,52,8(2014) 5005-5013) proposed a high-resolution inverse synthetic aperture radar ISAR imaging method based on sparse Bayesian. This method first performs distance alignment on the original echoes, and then self-focusses. Then, in order to perform Bayesian analysis, sparse Bayesian learning adopts a hierarchical prior structure, and obtains all corresponding parameters through a confidence-maximization program without artificial Intervention, and finally output the final inverse synthetic aperture radar ISAR image. The disadvantage of this method is that this method does not deal with the noise in the echo data, which leads to poor target focusing effect, especially when the signal-to-noise ratio is very low, the robustness of this method to noise is not good. Not very good.

The patent application "Inverse SAR method for maneuvering targets based on sparse aperture" (application number: 2014101401235, publication number: CN103901429A) filed by Xidian University discloses a method for inverse synthetic aperture radar imaging for maneuvering targets based on sparse apertures. The method uses an improved eigenvector self-focusing method to achieve accurate phase correction, constructs an objective function for the phase-compensated echo signal, and reconstructs the precise motion-compensated sparse aperture signal into a full aperture using an orthogonal matching pursuit algorithm. Finally, the high-resolution inverse synthetic aperture radar (ISAR) imaging is realized by performing fast Fourier transform on the echo signal of the sparse aperture. The shortcomings of this method are: the phase compensation of this method is not thorough enough, and the azimuth resolution is not high. Moreover, performing fast Fourier transform directly on the full-aperture signal can only reduce the noise amplitude of the background signal and cannot be all zero.

Contents of the invention

Aiming at the problem of low imaging resolution of inverse synthetic aperture radar in the above prior art, the present invention proposes a high-resolution imaging method of inverse synthetic aperture radar based on multi-layer co-sparseness, through analytic operator learning and signal denoising, so that Improved resolution and robustness of inverse synthetic aperture radar imaging.

The present invention comprises the steps:

(1) Input echo data:

Input the inverse synthetic aperture radar echo data after motion compensation and range compression;

(2) Analytical operator learning:

(2a) Initialize the number of iterations h of analytical operator learning to 1, and the value range of h is h∈[1,H _max ], where ∈ means belonging to a symbol, and H _max means the maximum number of iterations;

(2b) Learning the analytical operator of the radar echo data corresponding to the current iteration number of analytical operator learning;

(2c) Judging whether the termination condition of the analytic operator learning process is satisfied, if so, execute step (3), otherwise, execute step (2b) after adding 1 to the number of iterations of analytic operator learning;

(3) Signal denoising:

(3a) Initialize the number of iterations q of signal denoising to 1, the value range of q is q∈[1, Q _max ], and Q _max represents the maximum number of iterations;

(3b) performing signal denoising on the radar echo data corresponding to the current iteration number of signal denoising;

(3c) judging whether the termination condition of signal denoising is satisfied, if so, then perform step (4), otherwise, add 1 to the number of iterations of signal denoising and perform step (3b);

(4) Obtain the optimal analytical operator and echo data:

(4a) Initialize the number of iterations k of the joint process of analytic operator learning and signal denoising to 1, the value range of k is k∈[1,K _max ], and K _max represents the maximum number of iterations;

(4b) According to the following formula, calculate the joint convergence value of the current iteration of analytic operator learning and signal denoising:

Among them, _Val1k represents the joint convergence value of the _k -th iteration of analytic operator learning and signal denoising, |||| , s _m represents the data of the mth column in the inverse synthetic aperture radar echo data, Indicates the denoised data of the data in the mth column in the inverse synthetic aperture radar echo data, λ indicates the Lagrangian multiplier, and ||·|| _F indicates the Frobenius norm operation;

(4c) Judging whether the conditions of the optimal analytical operator and echo data are satisfied, if so, execute step (5), otherwise, increase the number of iterations of the joint process of analytical operator learning and signal denoising and then execute step ( 2);

(5) Restoring low-resolution images:

According to the learned analytic operator and the denoised inverse synthetic aperture radar data, the modified orthogonal matching pursuit OMP algorithm is used to reconstruct the low-resolution inverse synthetic aperture radar ISAR image;

(6) Judging whether the current imaging layer number is the maximum number of layers P, if so, then perform step (8), otherwise, perform step (7);

(7) Convert the low-resolution ISAR image to the data domain by inverse Fourier transform, and perform step (2);

(8) Output the final high-resolution inverse synthetic aperture radar ISAR image.

Compared with the prior art, the present invention has the following advantages:

First, the present invention overcomes the problem of inaccurate phase compensation of echo data in the prior art, resulting in low azimuth resolution, by learning the analytical operator of each column of data in the radar echo data, making the present invention The invention improves the azimuth resolution in inverse synthetic aperture radar imaging.

Second, in the present invention, by performing signal denoising on each column of data in the radar echo data, the problem of large background noise in inverse synthetic aperture radar imaging using echo data with a low signal-to-noise ratio in the prior art is overcome, making the present invention The invention improves the robustness of inverse synthetic aperture radar imaging to noise.

Description of drawings

Fig. 1 is a flowchart of the present invention;

Fig. 2 is a graph of the convergence of the analytical operator in the iterative learning of the analytical operator of the present invention;

Fig. 3 is a graph of signal denoising convergence in signal denoising iterative learning of the present invention;

Fig. 4 is a graph of the joint optimization convergence of analytical operator learning and signal denoising in the present invention;

FIG. 5 is an imaging diagram of actual measurement data with a signal-to-noise ratio of -4dB and 32 pulses by the present invention and the existing inverse synthetic aperture radar imaging method.

FIG. 6 is an imaging diagram of actual measurement data of 64 pulses with a signal-to-noise ratio of -4 dB in the present invention and the existing inverse synthetic aperture radar imaging method.

detailed description

The present invention will be described in further detail below in conjunction with the accompanying drawings.

With reference to accompanying drawing 1, the specific steps of the present invention are described as follows.

Step 1, input echo data.

Input motion compensated and range compressed inverse synthetic aperture radar echo data.

Step 2, analytical operator learning.

Initialize the number of iterations h of analytical operator learning to 1, and the value range of h is h∈[1,H _max ], where ∈ means belonging to a symbol, and H _max means the maximum number of iterations.

The analytical operator of the radar echo data corresponding to the current iteration number of the analytical operator learning is learned, and the steps of the learning process of the analytical operator of the radar echo data are as follows:

Step 1, the analytical operator of the current column of inverse synthetic aperture radar Initialized as a partial Fourier matrix F;

Step 2, according to the following formula, calculate the data in the image domain of the current column data in the inverse synthetic aperture radar echo data:

in, represents the image domain data after the hth iteration of the mth column of the inverse synthetic aperture radar ISAR image, Represents the analytic operator obtained in the hth iteration of the mth column of the inverse synthetic aperture radar ISAR echo data;

Step 3, according to the following formula, update the image domain data of the current column of the inverse synthetic aperture radar ISAR image:

in, Indicates the value of column m and row n of the updated inverse synthetic aperture radar ISAR image, means not belonging to the symbol, Λ means middle front The subscript set of the minimum value;

Step 4, according to the following formula, calculate the analytical operator of the current column data of the inverse synthetic aperture radar ISAR echo data The gradient of:

Among them, _Fg represents the analytical operator of the mth column data of the inverse synthetic aperture radar ISAR echo data The gradient of , e represents the image domain data of the mth column of the inverse synthetic aperture radar image and the image domain data of the mth column of the updated inverse SAR image The error between, T represents the transpose operation;

Step 5, according to the following formula, update the analytical operator of the current column data of the inverse synthetic aperture radar ISAR echo data:

in, Indicates the analytical operator of the updated inverse synthetic aperture radar ISAR echo data column m data, η indicates the learning rate of the analytical operator of the inverse synthetic aperture radar ISAR echo data column m data, the value range of η is η ∈(0,1);

Step 6: Calculate the current convergence value of the data analysis operator in the current column of inverse synthetic aperture radar ISAR echo data according to the following formula:

Among them, Val2 ^h represents the current convergence value of the data analysis operator in the mth column of the inverse synthetic aperture radar ISAR echo data.

Judging whether the termination condition of the analytic operator learning process is satisfied, if so, then execute step (3), otherwise, perform step (2b) after adding 1 to the number of iterations of the analytic operator learning, the termination condition of the analytic operator learning is means one of the following conditions:

In the first case, the joint convergence value is less than 0.01;

In the second case, the number of iterations of the analytic operator learning process reaches 20 times.

Referring to accompanying drawing 2, the effect of the analysis operator optimization process of the present invention will be further described. Fig. 2(a) is a graph of the convergence of the analytical operator for the first iteration. The abscissa in Fig. 2(a) represents the number of iterations, and the ordinate represents the convergence value of the analytical operator. Fig. 2(b) is a graph showing the convergence of the analytical operator in the second iteration. The abscissa in Fig. 2(b) represents the number of iterations, and the ordinate represents the convergence value of the analytical operator. Fig. 2(c) is a graph showing the convergence of the analytical operator in the third iteration. The abscissa in Fig. 2(c) represents the number of iterations, and the ordinate represents the convergence value of the analytical operator. Fig. 2(d) is a graph showing the convergence of the analytical operator in the fourth iteration. The abscissa in Fig. 2(d) represents the number of iterations, and the ordinate represents the convergence value of the analytical operator. It can be seen from Fig. 2(a) that the analytic operator convergence value Val2 ^h gradually decreases during the first iteration of the analytic operator, which shows that the analytic operator optimization process of the present invention is effective. The analytical operator convergence value Val2 ^h in the curves of Fig. 2 (b), (c) and (d) has the same downward trend. The variation of the analytical operator convergence value Val2 ^h in each figure also decreases after the same number of iterations, which shows that our analytical operator convergence value Val2 ^h will eventually converge.

Step 3, signal denoising.

Initialize the number of iterations q of signal denoising to 1, the value range of q is q∈[1,Q _max ], and Q _max represents the maximum number of iterations.

Performing signal denoising on the radar echo data corresponding to the current iteration number of signal denoising, the steps of performing the signal denoising process on the radar echo data are as follows:

In the first step, an optimization model for denoising the data signal of each column of inverse synthetic aperture radar ISAR echo data is established according to the following formula:

Among them, min( ) represents the minimum value operation, and s.t. represents the conditional constraint symbol;

In the second step, according to the following formula, the signal denoising solution model of each column of inverse synthetic aperture radar ISAR echo data is established:

Among them, ψ represents the value to be solved, and b represents a dual parameter whose value range is b∈(0,1), representing a constant whose value range is

Step 3, calculate the value of the soft threshold operation according to the following formula:

Among them, α means The value of , β means The value of ≥ means greater than less than the sign

Step 4, according to the following three formulas, the echo data of the mth column of the inverse synthetic aperture radar ISAR echo data Inverse synthetic aperture radar ISAR image column m image domain data Update the dual parameter b ^q :

in, Indicates the mth column data of the updated inverse synthetic aperture radar ISAR echo data, Represents the mth column data of the updated inverse synthetic aperture radar ISAR image domain data, and b ^q+1 represents the updated dual parameter;

Step 5, according to the following formula, calculate the current convergence value of data signal denoising in each column of inverse synthetic aperture radar ISAR echo data:

Among them, Val3 ^q represents the convergence value of data signal denoising in each column of inverse synthetic aperture radar ISAR echo data.

Judging whether the termination condition of signal denoising is satisfied, if so, then perform step (4), otherwise, add 1 to the number of iterations of signal denoising and perform step (3b), and the termination condition of signal denoising refers to the following conditions One of the cases:

In the first case, the joint convergence value is less than 0.01;

In the second case, the number of iterations of the signaling process reaches 30.

Referring to Fig. 3, the effect of the signal denoising optimization process of the present invention will be further described. Fig. 3(a) is a graph showing the signal denoising convergence of the first iteration, the abscissa in Fig. 3(a) represents the number of iterations, and the ordinate represents the convergence value of the signal denoising. Fig. 3(b) is a graph showing the signal denoising convergence of the second iteration, the abscissa in Fig. 3(b) represents the number of iterations, and the ordinate represents the signal denoising convergence value. Fig. 3(c) is a graph showing the convergence of the signal denoising in the third iteration, the abscissa in Fig. 3(c) represents the number of iterations, and the ordinate represents the convergence value of the signal denoising. FIG. 3( d ) is a graph showing the convergence of the signal denoising in the fourth iteration. The abscissa in FIG. 3( d ) represents the number of iterations, and the ordinate represents the convergence value of the signal denoising. It can be seen from each picture in Figure 3 that the convergence value Val3 ^q of signal denoising is increased first and then converges to a value. We put these figures together and we can find that the convergence value Val3 ^q of signal denoising is gradually decreases, which shows that the signal denoising method of the present invention is effective.

Step 4, obtaining the optimal analytical operator and echo data.

Initialize the number of iterations k of the joint process of analytic operator learning and signal denoising to 1, and the value range of k is k∈[1,K _max ], where K _max represents the maximum number of iterations.

Calculate the joint convergence value of the current iteration of analytic operator learning and signal denoising according to the following formula:

Among them, _Val1k represents the joint convergence value of the _k -th iteration of analytic operator learning and signal denoising, |||| , s _m represents the data of the mth column in the inverse synthetic aperture radar echo data, Indicates the denoised data of the mth column in the inverse SAR echo data, λ indicates the Lagrangian multiplier, and ||·|| _F indicates the Frobenius norm operation.

Judging whether the conditions of the optimal analytical operator and echo data are satisfied, if so, execute step (5), otherwise, perform step (2) after adding 1 to the iteration number of the joint process of analytical operator learning and signal denoising, The conditions for the optimal analytical operator and echo data refer to one of the following conditions:

In the first case, the joint convergence value is less than 0.01;

In the second case, the number of iterations of the joint process of analytic operator learning and signal denoising reaches 50.

Referring to Fig. 4, the effect of the joint optimization process of analytic operator learning and signal denoising in the present invention will be further described. The abscissa in Figure 4 represents the number of iterations, and the ordinate represents the joint convergence value of the analysis operator learning and signal denoising. It can be seen from Fig. 4 that the joint convergence value of the analytic operator learning and signal denoising is gradually reduced as the number of iterations increases, which shows that the process of analytic operator learning and signal denoising in the present invention is effective.

Step 5, restore the low-resolution image.

According to the obtained analytic operator and the denoised inverse synthetic aperture radar data, the modified orthogonal matching pursuit OMP algorithm is used to reconstruct the low-resolution inverse synthetic aperture radar ISAR image. The modified orthogonal matching pursuit The steps of OMP algorithm to reconstruct the low-resolution inverse synthetic aperture radar ISAR image are as follows:

In the first step, the number of iterations t of image restoration is initialized to 1, and the value range of t is t∈[1,T _max ], where T _max represents the maximum number of iterations of image restoration, Among them, N represents the number of discrete points in the azimuth direction of the high-resolution image, Indicates the co-sparsity.

In the second step, the data in the image domain is calculated according to the following formula:

a ^t = F _m r ^t

Among them, F _m represents the analytical operator, r ^t represents the difference value of the echo data of the tth iteration, which is initialized as

Step 3, find the position λ _t of the maximum value of |a ^t | according to the following formula:

Step 4: Update the position set Θ ^t , the analytical operator Φ ^t and the image domain data temporary solution a _ps according to the following formula:

Θ ^t = Θ ^t-1 ∪{λ _t }

Among them, Θ ^t represents the updated position set, which is initialized as represents the empty set symbol, ∪ represents the union symbol, denote the _λt -th row of F _m , and a _ps denote the temporary solution of the image domain data.

Step 5, update the difference according to the following formula:

r ^t+1 represents the difference between the updated echo data, and (·) ^-1 represents the inverse operation.

Step 6, get the final image domain data according to the following formula:

Among them, a _m represents the mth column data of the final image.

Step 6, judging whether the current number of imaging layers is the maximum number of layers P, if so, execute step (8), otherwise, execute step (7), the value of the maximum number of layers P is a positive integer greater than 1.

Step 7: Transform the low-resolution inverse synthetic aperture radar ISAR image into the data domain through inverse Fourier transform, and perform step (2).

Step 8, outputting the final high-resolution inverse synthetic aperture radar ISAR image.

The effects of the present invention will be further described below in combination with simulation experiments.

1. Simulation conditions:

The emulated operating system of the present invention is Windows7 ultimate edition 64-bit SP1 operating system, CPU is Intel (R) Core (TM) i5-2410M@2.30GHz, emulation software is in MATLAB R2014a.

2. Simulation content:

The emulation experiment of the present invention is, adopt the sparse Bayesian inverse synthetic aperture radar imaging method of the present invention and prior art, be respectively to signal-to-noise ratio be-4dB, pulse number be that 32 ship data and signal-to-noise ratio are- 4dB, the number of pulses is 64 ship data for simulation. The parameters of the radar system used in the present invention are as follows: the carrier frequency is 9.25GHz, the signal bandwidth is 500MHz, the pulse repetition frequency is 200Hz, and the total number of pulses is 256.

3. Simulation result analysis:

Referring to FIG. 5 , the present invention and the imaging diagram of the existing inverse synthetic aperture radar imaging technology will be further described. FIG. 5 shows the ISAR imaging diagram of the present invention and the existing inverse synthetic aperture radar imaging technology with a signal-to-noise ratio of -4dB and a pulse number of 32. The abscissa in Fig. 5 represents the number of range units, and the ordinate represents the number of Doppler units. Fig. 5(a) is an ISAR image obtained by processing the prior art inverse SAR technology. Fig. 5(b) is the inverse synthetic aperture radar ISAR imaging image processed by the method of the present invention. As can be seen from Fig. 5, when the number of pulses is very low, the resolution of the inverse synthetic aperture radar ISAR image obtained by the present invention is still very high, and the target will not be deformed due to the reduction of the number of pulses, so the present invention is comparable to the prior art. than has better resolution.

Referring to FIG. 6 , the imaging diagrams of the present invention and the existing inverse synthetic aperture radar imaging technology are further described. FIG. 6 shows the ISAR imaging diagrams of the present invention and the prior art when the signal-to-noise ratio is -4dB and the number of pulses is 64. The abscissa of Fig. 6 represents the number of range units, and the ordinate represents the number of Doppler units, wherein Fig. 6 (a) is an inverse synthetic aperture radar (ISAR) imaging map obtained by processing the prior art inverse synthetic aperture radar technology, Fig. 6 (b) is the inverse synthetic aperture radar (ISAR) imaging image processed by the method of the present invention. It can be seen from Fig. 6 that when the SNR of the ISAR image obtained by the present invention is very low, there is still little noise in the background, while there are a large number of noises in the ISAR image of the processing result of the prior art. Noise, so the present invention has stronger robustness compared with the prior art.

Claims (8)

1. A method for inverse synthetic aperture radar high-resolution imaging based on multi-layer co-sparse, comprising the steps of: (1) Input echo data: Input the inverse synthetic aperture radar echo data after motion compensation and range compression; (2) Analytical operator learning: (2a) Initialize the number of iterations h of analytical operator learning to 1, and the value range of h is h∈[1,H _max ], where ∈ means belonging to a symbol, and H _max means the maximum number of iterations; (2b) Learning the analytical operator of the radar echo data corresponding to the current iteration number of analytical operator learning; (2c) Judging whether the termination condition of the analytic operator learning process is satisfied, if so, execute step (3), otherwise, execute step (2b) after adding 1 to the number of iterations of analytic operator learning; (3) Signal denoising: (3a) Initialize the number of iterations q of signal denoising to 1, the value range of q is q∈[1, Q _max ], and Q _max represents the maximum number of iterations; (3b) performing signal denoising on the radar echo data corresponding to the current iteration number of signal denoising; (3c) judging whether the termination condition of signal denoising is satisfied, if so, then perform step (4), otherwise, add 1 to the number of iterations of signal denoising and perform step (3b); (4) Obtain the optimal analytical operator and echo data: (4a) Initialize the number of iterations k of the joint process of analytic operator learning and signal denoising to 1, the value range of k is k∈[1,K _max ], and K _max represents the maximum number of iterations; (4b) According to the following formula, calculate the joint convergence value of the current iteration of analytic operator learning and signal denoising: <mrow> <mi>V</mi> <mi>a</mi> <mi>l</mi> <msup> <mn>1</mn> <mi>k</mi> </msup> <mo>=</mo> <mo>|</mo> <mo>|</mo> <msub> <mi>F</mi> <mi>m</mi> </msub> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>|</mo> <msub> <mo>|</mo> <mn>1</mn> </msub> <mo>+</mo> <mfrac> <mi>&amp;lambda;</mi> <mn>2</mn> </mfrac> <mo>|</mo> <mo>|</mo> <msub> <mi>s</mi> <mi>m</mi> </msub> <mo>-</mo> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>|</mo> <msubsup> <mo>|</mo> <mi>F</mi> <mn>2</mn> </msubsup> </mrow> Among them, Val1 ^k represents the joint convergence value of the k-th iteration of analytic operator learning and signal denoising, || · || ₁ represents the 1-norm operation, and F _m represents the analytical operator, s _m represents the data of column m in the inverse synthetic aperture radar echo data, Indicates the denoised data of the data in the mth column in the inverse synthetic aperture radar echo data, λ indicates the Lagrangian multiplier, and ||·|| _F indicates the Frobenius norm operation; (4c) Judging whether the conditions of the optimal analytical operator and echo data are satisfied, if so, execute step (5), otherwise, increase the number of iterations of the joint process of analytical operator learning and signal denoising and then execute step ( 2); (5) Restoring low-resolution images: According to the learned analytic operator and the denoised inverse synthetic aperture radar data, the modified orthogonal matching pursuit OMP algorithm is used to reconstruct the low-resolution inverse synthetic aperture radar ISAR image; (6) Judging whether the current imaging layer number is the maximum number of layers P, if so, then perform step (8), otherwise, perform step (7); (7) Convert the low-resolution ISAR image to the data domain by inverse Fourier transform, and perform step (2); (8) Output the final high-resolution inverse synthetic aperture radar ISAR image. 2. the inverse synthetic aperture radar high-resolution imaging method based on multi-layer co-sparse according to claim 1, is characterized in that, the analytic operator of the radar echo data described in step (2b) carries out the step of learning process as follows : Step 1, the analytical operator of the current column of inverse synthetic aperture radar Initialized as a partial Fourier matrix F; Step 2, according to the following formula, calculate the data in the image domain of the current column data in the inverse synthetic aperture radar echo data: <mrow> <msubsup> <mi>a</mi> <mi>m</mi> <mi>h</mi> </msubsup> <mo>=</mo> <msubsup> <mi>F</mi> <mi>m</mi> <mi>h</mi> </msubsup> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> </mrow> 1 in, represents the image domain data after the hth iteration of the mth column of the inverse synthetic aperture radar ISAR image, Represents the analytic operator obtained in the hth iteration of the mth column of the inverse synthetic aperture radar ISAR echo data; Step 3, according to the following formula, update the image domain data of the current column of the inverse synthetic aperture radar ISAR image: <mrow> <msubsup> <mover> <mi>a</mi> <mo>^</mo> </mover> <mrow> <mi>n</mi> <mi>m</mi> </mrow> <mi>h</mi> </msubsup> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mn>0</mn> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <mi>n</mi> <mo>&amp;Element;</mo> <mi>&amp;Lambda;</mi> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msubsup> <mi>a</mi> <mrow> <mi>n</mi> <mi>m</mi> </mrow> <mi>h</mi> </msubsup> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <mi>n</mi> <mo>&amp;NotElement;</mo> <mi>&amp;Lambda;</mi> </mrow> </mtd> </mtr> </mtable> </mfenced> </mrow> in, Indicates the value of column m and row n of the updated inverse synthetic aperture radar ISAR image, means not belonging to the symbol, Λ means middle front The subscript set of the minimum value; Step 4, according to the following formula, calculate the analytical operator of the current column data of the inverse synthetic aperture radar ISAR echo data The gradient of: <mrow> <msub> <mi>F</mi> <mi>g</mi> </msub> <mo>=</mo> <mi>e</mi> <msup> <mrow> <mo>(</mo> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>)</mo> </mrow> <mi>T</mi> </msup> </mrow> Among them, _Fg represents the analytical operator of the mth column data of the inverse synthetic aperture radar ISAR echo data The gradient of , e represents the image domain data of the mth column of the inverse synthetic aperture radar image and the image domain data of the mth column of the updated inverse SAR image The error between, T represents the transpose operation; Step 5, according to the following formula, update the analytical operator of the current column data of the inverse synthetic aperture radar ISAR echo data: <mrow> <msubsup> <mi>F</mi> <mi>m</mi> <mrow> <mi>h</mi> <mo>+</mo> <mn>1</mn> </mrow> </msubsup> <mo>=</mo> <msubsup> <mi>F</mi> <mi>m</mi> <mi>h</mi> </msubsup> <mo>-</mo> <msub> <mi>&amp;eta;F</mi> <mi>g</mi> </msub> </mrow> in, Indicates the analytical operator of the updated inverse synthetic aperture radar ISAR echo data column m data, η indicates the learning rate of the analytical operator of the inverse synthetic aperture radar ISAR echo data column m data, the value range of η is η ∈(0,1); Step 6: Calculate the current convergence value of the data analysis operator in the current column of inverse synthetic aperture radar ISAR echo data according to the following formula: <mrow> <mi>V</mi> <mi>a</mi> <mi>l</mi> <msup> <mn>2</mn> <mi>h</mi> </msup> <mo>=</mo> <mo>|</mo> <mo>|</mo> <msubsup> <mi>F</mi> <mi>m</mi> <mi>h</mi> </msubsup> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>|</mo> <msub> <mo>|</mo> <mn>1</mn> </msub> </mrow> Among them, Val2 ^h represents the current convergence value of the data analysis operator in the mth column of the inverse synthetic aperture radar ISAR echo data. 3. the inverse synthetic aperture radar high-resolution imaging method based on multi-layer co-sparse according to claim 1, is characterized in that, the termination condition of the analytical operator learning described in step (2c) refers to in the following conditions One situation: In the first case, the joint convergence value is less than 0.01; In the second case, the number of iterations of the analytic operator learning process reaches 20 times. 4. the inverse synthetic aperture radar high-resolution imaging method based on multi-layer co-sparse according to claim 1, is characterized in that, the step of carrying out signal denoising process to radar echo data described in step (3b) is as follows: In the first step, an optimization model for denoising the data signal of each column of inverse synthetic aperture radar ISAR echo data is established according to the following formula: <mrow> <munder> <mi>min</mi> <mrow> <msub> <mover> <mi>a</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>,</mo> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> </mrow> </munder> <mo>|</mo> <mo>|</mo> <msub> <mover> <mi>a</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>|</mo> <msub> <mo>|</mo> <mn>1</mn> </msub> <mo>+</mo> <mfrac> <mi>&amp;lambda;</mi> <mn>2</mn> </mfrac> <mo>|</mo> <mo>|</mo> <msub> <mi>s</mi> <mi>m</mi> </msub> <mo>-</mo> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>|</mo> <msub> <mo>|</mo> <mi>F</mi> </msub> <mo>,</mo> <mi>s</mi> <mo>.</mo> <mi>t</mi> <mo>.</mo> <msub> <mover> <mi>a</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>=</mo> <msub> <mi>F</mi> <mi>m</mi> </msub> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> </mrow> Among them, min( ) represents the minimum value operation, and s.t. represents the conditional constraint symbol; In the second step, according to the following formula, the signal denoising solution model of each column of inverse synthetic aperture radar ISAR echo data is established: <mrow> <mi>&amp;psi;</mi> <mo>=</mo> <mo>|</mo> <mo>|</mo> <msub> <mover> <mi>a</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>|</mo> <msub> <mo>|</mo> <mn>1</mn> </msub> <mo>+</mo> <mfrac> <mi>&amp;lambda;</mi> <mn>2</mn> </mfrac> <mo>|</mo> <mo>|</mo> <msub> <mi>s</mi> <mi>m</mi> </msub> <mo>-</mo> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>|</mo> <msubsup> <mo>|</mo> <mi>F</mi> <mn>2</mn> </msubsup> <mo>+</mo> <mfrac> <mi>&amp;gamma;</mi> <mn>2</mn> </mfrac> <mo>|</mo> <mo>|</mo> <mi>b</mi> <mo>+</mo> <msub> <mi>F</mi> <mi>m</mi> </msub> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>-</mo> <msub> <mover> <mi>a</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>|</mo> <msubsup> <mo>|</mo> <mi>F</mi> <mn>2</mn> </msubsup> <mo>-</mo> <mfrac> <mi>&amp;gamma;</mi> <mn>2</mn> </mfrac> <mo>|</mo> <mo>|</mo> <mi>b</mi> <mo>|</mo> <msubsup> <mo>|</mo> <mi>F</mi> <mn>2</mn> </msubsup> </mrow> Among them, ψ represents the value to be solved, and b represents a dual parameter whose value range is b∈(0,1), representing a constant whose value range is Step 3, calculate the value of the soft threshold operation according to the following formula: <mrow> <msub> <mi>f</mi> <mi>&amp;alpha;</mi> </msub> <mrow> <mo>(</mo> <mi>&amp;beta;</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mi>&amp;beta;</mi> <mo>-</mo> <mi>&amp;alpha;</mi> <mi>sgn</mi> <mrow> <mo>(</mo> <mi>&amp;beta;</mi> <mo>)</mo> </mrow> </mrow> </mtd> <mtd> <mrow> <mo>|</mo> <mi>&amp;beta;</mi> <mo>|</mo> <mo>&amp;GreaterEqual;</mo> <mi>&amp;alpha;</mi> </mrow> </mtd> </mtr> <mtr> <mtd> <mn>0</mn> </mtd> <mtd> <mrow> <mi>o</mi> <mi>t</mi> <mi>h</mi> <mi>e</mi> <mi>r</mi> <mi>w</mi> <mi>i</mi> <mi>s</mi> <mi>e</mi> </mrow> </mtd> </mtr> </mtable> </mfenced> </mrow> Among them, α means The value of , β means The value of ≥ means greater than less than the sign Step 4, according to the following three formulas, the echo data of the mth column of the inverse synthetic aperture radar ISAR echo data Inverse synthetic aperture radar ISAR image column m image domain data Update the dual parameter b ^q : <mrow> <msubsup> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> <mrow> <mi>q</mi> <mo>+</mo> <mn>1</mn> </mrow> </msubsup> <mo>=</mo> <mfrac> <mn>1</mn> <mrow> <mi>&amp;lambda;</mi> <mo>+</mo> <mi>&amp;gamma;</mi> </mrow> </mfrac> <mo>&amp;lsqb;</mo> <mi>&amp;lambda;</mi> <msubsup> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> <mi>q</mi> </msubsup> <mo>+</mo> <msub> <mi>&amp;gamma;F</mi> <mi>m</mi> </msub> <mrow> <mo>(</mo> <msubsup> <mover> <mi>a</mi> <mo>^</mo> </mover> <mi>m</mi> <mi>q</mi> </msubsup> <mo>-</mo> <msup> <mi>b</mi> <mi>q</mi> </msup> <mo>)</mo> </mrow> <mo>&amp;rsqb;</mo> </mrow> <mrow> <msubsup> <mover> <mi>a</mi> <mo>^</mo> </mover> <mi>m</mi> <mrow> <mi>q</mi> <mo>+</mo> <mn>1</mn> </mrow> </msubsup> <mo>=</mo> <msub> <mi>f</mi> <mfrac> <mn>1</mn> <mi>&amp;gamma;</mi> </mfrac> </msub> <mo>{</mo> <msub> <mi>F</mi> <mi>m</mi> </msub> <msubsup> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> <mrow> <mi>q</mi> <mo>+</mo> <mn>1</mn> </mrow> </msubsup> <mo>+</mo> <msup> <mi>b</mi> <mi>q</mi> </msup> <mo>}</mo> </mrow> <mrow> <msup> <mi>b</mi> <mrow> <mi>q</mi> <mo>+</mo> <mn>1</mn> </mrow> </msup> <mo>=</mo> <msup> <mi>b</mi> <mi>q</mi> </msup> <mo>+</mo> <mrow> <mo>(</mo> <msub> <mi>F</mi> <mi>m</mi> </msub> <msubsup> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> <mrow> <mi>q</mi> <mo>+</mo> <mn>1</mn> </mrow> </msubsup> <mo>-</mo> <msubsup> <mover> <mi>a</mi> <mo>^</mo> </mover> <mi>m</mi> <mrow> <mi>q</mi> <mo>+</mo> <mn>1</mn> </mrow> </msubsup> <mo>)</mo> </mrow> </mrow> in, Indicates the mth column data of the updated inverse synthetic aperture radar ISAR echo data, Represents the mth column data of the updated inverse synthetic aperture radar ISAR image domain data, and b ^q+1 represents the updated dual parameter; Step 5, according to the following formula, calculate the current convergence value of data signal denoising in each column of inverse synthetic aperture radar ISAR echo data: <mrow> <mi>V</mi> <mi>a</mi> <mi>l</mi> <msup> <mn>3</mn> <mi>q</mi> </msup> <mo>=</mo> <mo>|</mo> <mo>|</mo> <msub> <mi>F</mi> <mi>m</mi> </msub> <msubsup> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> <mi>q</mi> </msubsup> <mo>|</mo> <msub> <mo>|</mo> <mn>1</mn> </msub> <mo>+</mo> <mfrac> <mi>&amp;lambda;</mi> <mn>2</mn> </mfrac> <mo>|</mo> <mo>|</mo> <msub> <mi>s</mi> <mi>m</mi> </msub> <mo>-</mo> <msubsup> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> <mi>q</mi> </msubsup> <mo>|</mo> <msubsup> <mo>|</mo> <mi>F</mi> <mn>2</mn> </msubsup> </mrow> Among them, Val3 ^q represents the convergence value of data signal denoising in each column of inverse synthetic aperture radar ISAR echo data. 5. the inverse synthetic aperture radar high-resolution imaging method based on multi-layer co-sparse according to claim 1, is characterized in that, the termination condition of the signal denoising described in step (3c) refers to one of the following conditions situation: In the first case, the joint convergence value is less than 0.01; In the second case, the number of iterations of the signaling process reaches 30. 6. the inverse synthetic aperture radar high-resolution imaging method based on multi-layer co-sparse according to claim 1, is characterized in that, the optimal analytical operator described in step (4c) and the condition of echo data refer to One of the following conditions: In the first case, the joint convergence value is less than 0.01; In the second case, the number of iterations of the joint process of analytic operator learning and signal denoising reaches 50. 7. the inverse synthetic aperture radar high-resolution imaging method based on multi-layer co-sparse according to claim 1, is characterized in that, utilizes the modified orthogonal matching pursuit OMP algorithm described in step (5) to reconstruct low-resolution The inverse synthetic aperture radar ISAR image process of high rate, according to the following steps: In the first step, the number of iterations t of image restoration is initialized to 1, and the value range of t is t∈[1,T _max ], where T _max represents the maximum number of iterations of image restoration, Among them, N represents the number of discrete points in the azimuth direction of the high-resolution image, Indicates the co-sparsity. In the second step, the data in the image domain is calculated according to the following formula: a ^t = F _m r ^t Among them, F _m represents the analytical operator, r ^t represents the difference value of the echo data of the tth iteration, which is initialized as Step 3, find the position λ _t of the maximum value of |a ^t | according to the following formula: <mrow> <msub> <mi>&amp;lambda;</mi> <mi>t</mi> </msub> <mo>=</mo> <munder> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> <mrow> <mi>n</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mi>N</mi> </mrow> </munder> <mo>|</mo> <msup> <mi>a</mi> <mi>t</mi> </msup> <mo>|</mo> </mrow> Step 4: Update the position set Θ ^t , the analytical operator Φ ^t and the image domain data temporary solution a _ps according to the following formula: Θ ^t = Θ ^t-1 ∪{λ _t } <mrow> <msup> <mi>&amp;Phi;</mi> <mi>t</mi> </msup> <mo>=</mo> <mo>&amp;lsqb;</mo> <msup> <mi>&amp;Phi;</mi> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> </msup> <mo>;</mo> <msub> <mi>f</mi> <msub> <mi>&amp;lambda;</mi> <mi>t</mi> </msub> </msub> <mo>&amp;rsqb;</mo> </mrow> <mrow> <msub> <mi>a</mi> <mrow> <mi>p</mi> <mi>s</mi> </mrow> </msub> <mo>=</mo> <msup> <mi>&amp;Phi;</mi> <mi>t</mi> </msup> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> </mrow> Among them, Θ ^t represents the updated position set, which is initialized as represents the empty set symbol, ∪ represents the union symbol, denote the _λt -th row of F _m , and a _ps denote the temporary solution of the image domain data. Step 5, update the difference according to the following formula: <mrow> <msup> <mi>r</mi> <mrow> <mi>t</mi> <mo>+</mo> <mn>1</mn> </mrow> </msup> <mo>=</mo> <msub> <mover> <mi>s</mi> <mo>^</mo> </mover> <mi>m</mi> </msub> <mo>-</mo> <msup> <mrow> <mo>(</mo> <msup> <mi>&amp;Phi;</mi> <mi>t</mi> </msup> <mo>)</mo> </mrow> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msup> <msub> <mi>a</mi> <mrow> <mi>p</mi> <mi>s</mi> </mrow> </msub> </mrow> r ^t+1 represents the difference between the updated echo data, and (·) ^-1 represents the inverse operation. Step 6, get the final image domain data according to the following formula: <mrow> <msub> <mi>a</mi> <mi>m</mi> </msub> <mrow> <mo>(</mo> <mi>n</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>a</mi> <mrow> <mi>p</mi> <mi>s</mi> </mrow> </msub> <mrow> <mo>(</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow> </mtd> <mtd> <mrow> <mi>n</mi> <mo>=</mo> <mi>&amp;Theta;</mi> <mrow> <mo>(</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> <mtr> <mtd> <mn>0</mn> </mtd> <mtd> <mrow> <mi>o</mi> <mi>t</mi> <mi>h</mi> <mi>e</mi> <mi>r</mi> <mi>w</mi> <mi>i</mi> <mi>s</mi> <mi>e</mi> </mrow> </mtd> </mtr> </mtable> </mfenced> </mrow> Among them, a _m represents the mth column data of the final image. 8. The inverse synthetic aperture radar high-resolution imaging method based on multi-layer co-sparse according to claim 1, wherein the value of the maximum number of layers P described in step (6) is a positive integer greater than 1.