Structured sparse based multi-channel radar forward-looking super-resolution imaging method

Abstract

The present application relates to a kind of based on structured sparse multi-channel radar forward-looking super-resolution imaging method, it is related to radar technical field, including: the echo signal model under the establishment of multi-channel radar forward-looking imaging mode, the target echo signal according to the echo signal model is processed, and the corrected target echo signal is obtained;According to the structured sparse characteristic analysis result of the structured sparse characteristic of existing image that the corrected target echo signal is obtained;Iterative convolution processing is carried out to the structured sparse characteristic analysis result, and the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging is obtained;The constraint optimization problem of multi-channel radar structured sparse forward-looking imaging is solved by alternating direction multiplier method, and the forward-looking high-resolution image of target is obtained.The above technical scheme is obtained by iterative convolution processing structured sparse characteristic analysis result, and the anti-noise capability is improved by iterative convolution, and finally the forward-looking high-resolution image with structured sparse characteristic is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar, in particular to a multi-channel radar forward-looking super-resolution imaging method based on structured sparsity. BACKGROUND

[0002] Radar forward-looking imaging can provide fine electromagnetic scattering characteristics of targets in the front of the moving platform, which is crucial in applications such as terrain measurement, precision guidance and autonomous driving. At present, the Doppler resolution principle relied on by the radar using synthetic aperture system is completely invalid when imaging the forward-looking area, forming a forward-looking "blind area". For the radar system using real beam imaging, the azimuth resolution is completely limited by the beam width, and the imaging resolution cannot meet the actual needs of the moving platform.

[0003] The imaging method based on compressive sensing utilizes the sparse characteristics of target distribution in the imaging scene to push the echo data of the short-aperture antenna to the echo data corresponding to the long-aperture antenna, providing feasibility for realizing radar forward-looking super-resolution imaging. However, the traditional imaging method based on compressive sensing (CS) often does not consider the relationship between different scattering points, and the image is composed of independent strong scattering points, leading to difficulties in target recognition based on the forward-looking image. In addition, noise can easily affect the imaging method based on compressive sensing, resulting in a large number of false scattering points in the imaging result, thereby affecting the overall imaging performance. SUMMARY

[0004] In order to solve the above problems existing in the prior art, the present application provides a multi-channel radar forward-looking super-resolution imaging method based on structured sparsity.

[0005] According to a first aspect of the embodiment of the present application, a multi-channel radar forward-looking super-resolution imaging method based on structured sparsity is provided, and the method comprises:

[0006] establishing an echo signal model in a multi-channel radar forward-looking imaging mode, processing target echo signals obtained according to the echo signal model to obtain corrected target echo signals;

[0007] obtaining a structured sparse characteristic analysis result corresponding to the corrected target echo signals according to the structured sparse characteristics of the existing image;

[0008] iteratively convolving the structured sparse characteristic analysis result to obtain a constraint optimization problem of multi-channel radar structured sparse forward-looking imaging;

[0009] solving the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging by an alternating direction multiplier method to obtain a forward-looking high-resolution image of the target.

[0010] Optionally, the method further comprises: establishing an echo signal model in a multi-channel radar forward-looking imaging mode; processing a target echo signal obtained according to the echo signal model to obtain a corrected target echo signal.

[0011] establishing an echo signal model in a multi-channel radar forward-looking imaging mode to obtain a target echo signal;

[0012] performing pulse compression on the target echo signal to obtain a pulse-compressed target echo signal;

[0013] performing range walk correction on the pulse-compressed target echo signal to obtain the corrected target echo signal.

[0014] Optionally, the method further comprises: performing iterative convolution processing on the structured sparse characteristic analysis result to obtain a constraint optimization problem of multi-channel radar structured sparse forward-looking imaging.

[0015] performing motion compensation on the corrected target echo signal to obtain a two-dimensional echo matrix;

[0016] performing iterative convolution processing on the structured sparse characteristic analysis result to obtain a convolution processing result;

[0017] establishing the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging according to the convolution processing result and the two-dimensional echo matrix.

[0018] Optionally, the method further comprises: converting the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging into a convex optimization problem.

[0019] converting the convex optimization problem into an expression form of an augmented Lagrangian function to obtain a converted convex optimization problem;

[0020] solving the converted convex optimization problem by the alternating direction multiplier method to obtain the forward-looking high-resolution image of the target.

[0021] Optionally, the method further comprises:

[0022] converting the convex optimization problem into an expression form of an augmented Lagrangian function to obtain a converted convex optimization problem;

[0023] solving the converted convex optimization problem by the alternating direction multiplier method to obtain the forward-looking high-resolution image of the target.

[0024] Optionally, the corrected target echo signal is represented as follows:

[0025]

[0026] wherein S rc denotes the corrected target echo signal, tau denotes fast time, t denotes slow time, A denotes the complex scattering coefficient of the target, c denotes the speed of light, B denotes the bandwidth, R0 denotes the slant range from the center of the array antenna of the multichannel radar to the target, exp(·) denotes the exponential function, j denotes the imaginary unit, R(t) denotes the distance from the nth antenna element of the multichannel radar to the target P, lambda denotes the wavelength of the multichannel radar, v a denotes the switching speed of the multichannel radar element transmitting signal, x0 denotes the horizontal coordinate of the target, y0 denotes the vertical coordinate of the target, and v denotes the flight speed of the radar platform of the multichannel radar.

[0027] Optionally, the two-dimensional echo matrix is represented as follows:

[0028] S rc = Y + E = FX + E.

[0029] wherein S rc denotes the two-dimensional echo matrix, Y denotes the signal matrix, E denotes the noise matrix, F is the dictionary matrix, and X is the two-dimensional image matrix of the radar image.

[0030] Optionally, the constraint optimization problem of the structured sparse forward-looking imaging of the multichannel radar is represented as follows:

[0031]

[0032] wherein omega denotes the convolution kernel matrix, ||·||1 denotes the l1 norm, ||·||F denotes the Frobenius norm, F denotes the two-dimensional convolution operation, denotes the matrix Hadamard product operation, epsilon denotes an arbitrary minimum value, and lambda' denotes a regularization parameter, denotes the optimization objective of the constraint optimization problem of the structured sparse forward-looking imaging of the multichannel radar, and s.t. rc = FX + E denotes the constraint condition of the constraint optimization problem of the structured sparse forward-looking imaging of the multichannel radar.

[0033] The technical scheme provided by the present application can include the following beneficial effects:

[0034] Through the above technical scheme, the present application proposes a structured sparse-based forward-looking super-resolution imaging method, obtains a forward-looking image with a structured characteristic of a target, improves the anti-noise capability of imaging through iterative convolution, and obtains a high-quality forward-looking high-resolution image. ​

[0035] Other features and advantages of the present application will be made clear to those skilled in the art from the following detailed description. BRIEF DESCRIPTION OF DRAWINGS

[0036] The accompanying drawings are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings:

[0037] Figure 1 is a flow chart of a structured sparse based multi-channel radar forward looking super-resolution imaging method according to an exemplary embodiment.

[0038] Figure 2 is a schematic diagram of the observation geometry of a multi-channel radar forward looking imaging according to an exemplary embodiment.

[0039] Figure 3a is an imaging result of a long aperture according to an exemplary embodiment.

[0040] Figure 3b is an imaging result of a short aperture according to an exemplary embodiment.

[0041] Figure 4 is a schematic diagram of a real test experimental scene according to an exemplary embodiment.

[0042] Figure 5 is a schematic diagram of the key parameters of a real test experimental radar system according to an exemplary embodiment.

[0043] Figure 6a is a schematic diagram of an imaging result of the present application according to an exemplary embodiment.

[0044] Figure 6b is a schematic diagram of an imaging result of a real beam imaging method according to an exemplary embodiment.

[0045] Figure 6c is a schematic diagram of an imaging result of a traditional compressive sensing based imaging method according to an exemplary embodiment.

[0046] Figure 6d is a schematic diagram of an imaging result of a weighted compressive sensing based imaging method according to an exemplary embodiment.

[0047] Figure 7a is a schematic diagram of yet another imaging result of the present application according to an exemplary embodiment.

[0048] Figure 7b is a schematic diagram of yet another imaging result of a real beam imaging method according to an exemplary embodiment.

[0049] Figure 7c is a schematic diagram of imaging results of yet another conventional compressive sensing based imaging method according to an example embodiment.

[0050] Figure 7d is a schematic diagram of imaging results of yet another weighted compressive sensing based imaging method according to an example embodiment. DETAILED DESCRIPTION

[0051] Figure 1 is a flowchart of a structured sparse based multichannel radar forward looking super-resolution imaging method according to an example embodiment, as shown in Figure 1 the method comprises the following steps.

[0052] S101, a model of echo signals in a multichannel radar forward looking imaging mode is established, and a target echo signal obtained according to the model of echo signals is processed to obtain a corrected target echo signal.

[0053] Optionally, S101 can comprise:

[0054] establishing a model of echo signals in a multichannel radar forward looking imaging mode to obtain a target echo signal;

[0055] pulse compression is performed on the target echo signal to obtain a pulse compressed target echo signal;

[0056] range walk correction is performed on the pulse compressed target echo signal to obtain a corrected target echo signal.

[0057] It can be understood that, Figure 2 is a schematic diagram of observation geometry of multichannel radar forward looking imaging according to an example embodiment, as shown in Figure 2 The multichannel radar works in a "single transmission and single reception" mode, that is, N array elements on the array antenna transmit signals and receive echoes in turn with a pulse repetition interval (PRI) as the time interval. Assuming that there is a target P(x0, y0, 0) in front of the radar platform, then the distance from the nth antenna element to the point target P is:

[0058]

[0059] where t represents the slow time, H and v are the height and flight speed of the radar platform respectively, and x0and y0represent the horizontal coordinate and vertical coordinate of the target P respectively. Assuming that d represents the element spacing, then v a = d / PRI is the switching speed of the element transmitting signals, that is, the equivalent azimuth motion speed.

[0060] In the forward imaging of small moving platforms such as unmanned aerial vehicles, the distance between the target and the radar platform is usually larger than the length of the array antenna, and the above formula can be approximated as:

[0061]

[0062] wherein, is the slant range from the center of the array antenna of the multi-channel radar to the target P.

[0063] Suppose that the linear frequency modulation signal s(τ) emitted by the radar is:

[0064]

[0065] wherein τ represents fast time, G represents a range window function, f c is the carrier frequency, and γ is the frequency modulation slope. If the complex scattering coefficient of the target P is A, then the echo signal s(τ, t) of the target after coherent demodulation is:

[0066]

[0067] wherein λ is the wavelength and c is the speed of light.

[0068] After pulse compression processing of the target echo signal s(τ, t), the target echo signal S rc (τ, t) after range direction pulse compression can be represented as:

[0069]

[0070] wherein B is the bandwidth, and sinc(·) represents a sinc function.

[0071] Alternatively, the corrected target echo signal can be represented as:

[0072]

[0073] wherein S rc (τ, t) represents the corrected target echo signal, τ represents fast time, t represents slow time, A represents the complex scattering coefficient of the target, c represents the speed of light, B represents the bandwidth, R0 is the slant range from the center of the array antenna of the multi-channel radar to the target, exp(·) represents an exponential function, j represents an imaginary unit, R(t) represents the distance from the nth antenna element of the multi-channel radar to the target P, λ represents the wavelength of the multi-channel radar, v a represents the switching speed of the signal emitted by the antenna element of the multi-channel radar, x0 represents the horizontal coordinate of the target, y0 represents the vertical coordinate of the target, and v represents the flight speed of the radar platform of the multi-channel radar.

[0074] S102, obtaining a structured sparsity analysis result corresponding to the corrected target echo signal according to a structured sparsity characteristic of the existing image.

[0075] S103, performing iterative convolution processing on the structured sparsity analysis result to obtain a constraint optimization problem of multi-channel radar structured sparse forward-looking imaging.

[0076] Optionally, S103 can include:

[0077] motion compensation is performed on the corrected target echo signal to obtain a two-dimensional echo matrix;

[0078] iterative convolution processing is performed on the structured sparsity analysis result to obtain a convolution processing result;

[0079] a constraint optimization problem of multi-channel radar structured sparse forward-looking imaging is established according to the convolution processing result and the two-dimensional echo matrix.

[0080] Optionally, the two-dimensional echo matrix is represented as follows:

[0081] S rc =Y+E=FX+E;

[0082] wherein S rc represents a two-dimensional echo matrix, Y represents a signal matrix, E represents a noise matrix, F is a dictionary matrix, and X is a two-dimensional image matrix of a radar image.

[0083] It can be understood that in the forward-looking super-resolution imaging technology based on compressed sensing, short aperture data is used to derive long aperture data, which requires the introduction of prior information to ensure the accuracy of image reconstruction. Sparse prior is a commonly used method, which uses l0 or l1 norm constraint to utilize the sparse distribution of the target in the imaging scene, so as to reconstruct a high-resolution image. However, the existing sparse imaging algorithm can usually only reconstruct isolated strong scatterers, and cannot reflect the mutual relationship between scatterers or identify weak scatterers. For example, Figure 3a is an imaging result of long aperture data according to an example embodiment, Figure 3b is an imaging result of short aperture data according to an example embodiment, as shown in Figure 3a and Figure 3b By comparing the imaging results of long aperture and short aperture data, it can be seen that the real beam imaging technology can display continuous scatter point distribution and show structured sparse characteristics.

[0084] Optionally, l1 minimization is performed by introducing the continuity of scatter points, i.e. the convolution reweighting of the neighborhood value obtained in the previous iteration, and the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging is represented as follows:

[0085]

[0086] where ω denotes a convolution kernel matrix, ||·||1 denotes an l1 norm, ||·||F denotes a Frobenius norm, F denotes a two-dimensional convolution operation, denotes a matrix Hadamard product operation, ε denotes an arbitrary minimum value, λ' denotes a regularization parameter, denotes an optimization objective of a constrained optimization problem of multi-channel radar structured sparse forward-looking imaging, s.t. rc denotes a constraint condition of the constrained optimization problem of multi-channel radar structured sparse forward-looking imaging.

[0087] S104, solving the constrained optimization problem of multi-channel radar structured sparse forward-looking imaging by an alternating direction multiplier method to obtain a forward-looking high-resolution image of the target.

[0088] Optionally, S104 can include:

[0089] converting the constrained optimization problem of multi-channel radar structured sparse forward-looking imaging into a convex optimization problem;

[0090] solving the convex optimization problem by an alternating direction multiplier method to obtain a forward-looking high-resolution image of the target.

[0091] Optionally, solving the convex optimization problem by an alternating direction multiplier method to obtain a forward-looking high-resolution image of the target includes:

[0092] converting the convex optimization problem into an expression form of an augmented Lagrangian function to obtain a converted convex optimization problem;

[0093] iteratively solving the converted convex optimization problem by an alternating direction multiplier method to obtain a forward-looking high-resolution image of the target.

[0094] In an implementation, the convex optimization problem can be expressed as:

[0095]

[0096] converting the convex optimization problem into an expression form of an augmented Lagrangian function, and letting J=X, the optimization problem in the form of the augmented Lagrangian function is:

[0097]

[0098] where Q1 and Q2 are Lagrange multiplier matrices, and u1 and u2 are penalty term coefficients. The alternating direction multiplier method is used to alternately estimate variables J, X, and E, that is, one variable is estimated while other variables remain unchanged.

[0099] ​Then update the iteration to solve the value of each variable, the kth iteration of variable J is:

[0100] J k+1 = max(0, R1 k ) + min(0, R2 k );

[0101] wherein, η is an arbitrary minimum value to avoid singular values in the operation process.

[0102] The kth iteration of variable X is:

[0103]

[0104] wherein, I is an identity matrix.

[0105] The kth iteration of variable E is:

[0106]

[0107] Through continuous iteration to solve J, X, E, until the termination condition ||S rc -FX k || F / ||S rc || F ≤10 -6 or reach the maximum number of iterations, eventually get the pre-structured sparse characteristics of high-resolution image.

[0108] In an embodiment, the present application is verified based on experimental data. The experimental conditions are as follows:

[0109] Figure 4 is a kind of experimental scene schematic diagram according to an example embodiment, as shown in Figure 4 , wherein the experimental data is obtained from AWR2243 cascade radar acquisition, wherein the experimental scene is located in a parking lot, and the experimental target is two cars. Figure 5 is a kind of experimental experimental radar system working key parameter schematic diagram according to an example embodiment.

[0110] Measurement content 1: verify the super-resolution imaging ability of the method of the present application. Figure 6a is a kind of imaging result schematic diagram of the present application according to an example embodiment, Figure 6b is a kind of imaging result schematic diagram of a kind of real beam imaging method according to an example embodiment, Figure 6c is a kind of imaging result schematic diagram of a kind of traditional compressive sensing imaging method according to an example embodiment, Figure 6dis a kind of imaging result schematic diagram of weighted compressive sensing imaging method according to an exemplary embodiment, and the imaging result of the present application method, real beam imaging method, traditional compressive sensing imaging method and weighted compressive sensing imaging method are compared with 8 times super-resolution of front imaging as benchmark.

[0111] The imaging method proposed in the present application can distinguish the two cars from the visual angle, and the outline is more obvious, the structure information of the target is retained, and satisfactory imaging result can be generated. Figure 6a It can be seen that the present application has higher super-resolution imaging capability, and can retain more information of the target. Figure 6b It can be seen that the two cars in the imaging result of the real beam method are fused together and cannot be distinguished. Figure 6c And Figure 6d It can be seen that the traditional compressive sensing method and the weighted compressive sensing method are also difficult to distinguish the two cars, and there are many false scattering points in the imaging result, which seriously affects the subsequent target detection.

[0112] Actual measurement content 2: verify the noise robustness of the present application method. Figure 7a is another imaging result schematic diagram of the present application according to an exemplary embodiment, Figure 7b is another imaging result schematic diagram of real beam imaging method according to an exemplary embodiment, Figure 7c is another imaging result schematic diagram of traditional compressive sensing imaging method according to an exemplary embodiment, Figure 7d is another imaging result schematic diagram of weighted compressive sensing imaging method according to an exemplary embodiment, and the imaging result of the present application method, real beam imaging method, traditional compressive sensing imaging method and weighted compressive sensing imaging method are compared with 5dB as benchmark.

[0113] The target of the present application method is still clear and distinguishable, and has strong anti-noise ability, as shown in Figure 7a From Figure 7b It can be seen that although the real beam imaging method is not sensitive to noise, the two cars in the imaging result are completely fused together and cannot be distinguished. Figure 7c And Figure 7d It can be seen that when the signal-to-noise ratio is low, the imaging target of the compressive sensing method and the weighted compressive sensing method is submerged by noise, and there are many false scattering points.

[0114] The present application aims at the problem that the prior compressive sensing based forward imaging method does not consider the connection between target scattering points, and proposes a multi-channel radar forward super-resolution imaging method based on structured sparsity, which improves the imaging resolution while preserving the structural characteristics of the target, and has strong noise robustness. The structured sparse characteristics of the target are obtained by calculating the weight of each pixel in the next iteration through the convolution of the neighbor values in the current solution in the alternating direction multiplier method solving process, a multi-channel radar structured sparse forward super-resolution imaging model based on convolution weighting is established, and the imaging performance is improved.

[0115] The preferred embodiments of the present application are described in detail above with reference to the drawings, but the present application is not limited to the specific details in the above-described embodiments, and various simple modifications can be made to the technical solutions of the present application within the technical concept of the present application, and these simple modifications all belong to the protection scope of the present application.

[0116] In addition, it should be noted that each specific technical feature described in the above specific embodiments can be combined in any appropriate manner without contradiction, and in order to avoid unnecessary repetition, the present application will not further describe various possible combinations.

[0117] In addition, various different embodiments of the present application can also be combined in any manner, as long as they do not deviate from the idea of the present application, and they should also be considered as disclosed by the present application.

Claims

1. A method for structured sparse based multichannel radar forward looking super resolution imaging, characterized in that, The method comprises: establishing an echo signal model in a multi-channel radar forward-looking imaging mode, processing a target echo signal obtained according to the echo signal model to obtain a corrected target echo signal; obtaining a structured sparse characteristic analysis result corresponding to the corrected target echo signal according to a structured sparse characteristic of an existing image; performing iterative convolution processing on the structured sparse characteristic analysis result to obtain a constraint optimization problem of multi-channel radar structured sparse forward-looking imaging; solving the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging by an alternating direction multiplier method to obtain a forward-looking high-resolution image of the target; wherein the iterative convolution processing on the structured sparse characteristic analysis result to obtain the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging comprises: performing motion compensation on the corrected target echo signal to obtain a two-dimensional echo matrix; performing iterative convolution processing on the structured sparse characteristic analysis result to obtain a convolution processing result; establishing the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging according to the convolution processing result and the two-dimensional echo matrix; the two-dimensional echo matrix is expressed as follows: ； wherein, denotes the two-dimensional echo matrix, denotes the signal matrix, denotes the noise matrix, is a dictionary matrix, is a two-dimensional image matrix of the radar image; the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging is expressed as follows: ； wherein, denotes a convolution kernel matrix, denotes norm, denotes the Frobenius norm, denotes a two-dimensional convolution operation, denotes a matrix Hadamard product operation, denotes an arbitrary minimum value, denotes a regularization parameter, denotes an optimization objective of the constrained optimization problem of the multi-channel radar structured sparse forward-looking imaging, denotes a constraint condition of the constrained optimization problem of the multi-channel radar structured sparse forward-looking imaging.

2. The method of claim 1, wherein the method is implemented by a radar system comprising a plurality of antennas, each antenna having a respective transmit-receive channel, and wherein the method is implemented by a processor of the radar system. the establishment of the echo signal model in the multi-channel radar forward-looking imaging mode and the processing of the target echo signal obtained according to the echo signal model to obtain the corrected target echo signal comprises: establishing an echo signal model in a multi-channel radar forward-looking imaging mode to obtain a target echo signal; performing pulse compression on the target echo signal to obtain a pulse-compressed target echo signal; performing range walk correction on the pulse-compressed target echo signal to obtain the corrected target echo signal.

3. The multi-channel radar forward-looking super-resolution imaging method based on structured sparse according to claim 1, characterized in that: the solving of the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging by the alternating direction multiplier method to obtain the forward-looking high-resolution image of the target comprises: converting the constraint optimization problem of multi-channel radar structured sparse forward-looking imaging into a convex optimization problem; solving the convex optimization problem by the alternating direction multiplier method to obtain the forward-looking high-resolution image of the target.

4. The method of claim 3, wherein the structured sparse based multichannel radar forward looking super resolution imaging is characterized by, the solving of the convex optimization problem by the alternating direction multiplier method to obtain the forward-looking high-resolution image of the target comprises: converting the convex optimization problem into an expression form of an augmented Lagrangian function to obtain a converted convex optimization problem; iteratively solving the converted convex optimization problem by the alternating direction multiplier method to obtain the forward-looking high-resolution image of the target.

5. The method of claim 2, wherein the method is implemented by a radar system comprising a plurality of antennas, each antenna having a respective transmit-receive channel, and wherein the method further comprises: transmitting a plurality of radar signals from the plurality of antennas; receiving a plurality of radar signals at the plurality of antennas; and processing the received plurality of radar signals to generate a radar image of the scene. the corrected target echo signal is expressed as follows: ； in, represents the corrected target echo signal, Indicates fast time, Indicates slow time, represents the complex scattering coefficient of the target, represents the speed of light, Indicates bandwidth, is the slant distance from the center of the array antenna of the multi-channel radar to the target, represents the exponential function, represents the imaginary unit, Indicates the first antenna elements to the target distance, represents the wavelength of the multi-channel radar, represents the switching speed of the array element transmission signal of the multi-channel radar, represents the horizontal coordinate of the target, represents the ordinate of the target, Indicates the radar platform flight speed of the multi-channel radar.

Images