Real aperture radar adaptive coherent imaging method based on regularization and angular super-resolution

Abstract

The application discloses a kind of real aperture radar adaptive combination regularization angle super-resolution imaging methods, first, the acquisition and preprocessing of echo data are carried out, regularization objective function is constructed, then by calculating normalized weighting matrix, and iteration initialization is carried out, weight factor is updated, finally, the target scattering coefficient is updated, and the azimuth super-resolution imaging result of entire echo matrix is obtained.The method of the application combines generalized sparse norm and generalized total variation norm as constraint term, while enhancing angle resolution and expanding target edge information, then using the iterative reweighting method based on data driving, the selection of regularization parameter is avoided, the number of manually selected parameters is reduced, compared with the existing combination norm method, it has stronger scale information reconstruction capability, and improves the scale reconstruction accuracy of real aperture radar on extended target.

Description

Technical Field

[0001] This invention belongs to the field of radar imaging technology, specifically relating to an adaptive combined regularized angle super-resolution imaging method for real aperture radar. Background Technology

[0002] The angular resolution of a real aperture radar is related to the antenna aperture. In practical applications, due to space constraints on the radar platform, the aperture size is limited, resulting in low angular resolution for real aperture radars, making it difficult to meet the application requirements for high-resolution imaging in fields such as autonomous landing of aircraft and terrain mapping.

[0003] Regularization is an effective way to improve the angular resolution of real aperture radar. The paper "Y. Zhao, JG Liu, B. Zhang, W. Hong, and Y.-R. Wu, 'Adaptive total variation regularization based sarimage despeckling and despeckling evaluation index,' IEEE Transactions on Geoscience and Remote Sensing, 2014, pp. 2765-2774" proposes an edge enhancement method based on the total variation norm, but the reconstruction performance of a single regularization term is limited. The paper "W. Huo, X. Tuo, Y. Zhang, Y. Zhang, and Y. Huang, 'Balanced tikhonov and total variation deconvolution approach for radar forward-looking super-resolution imaging,' IEEE Geoscience and Remote Sensing Letters, 2021, pp. 1-5" proposes a combined regularization method incorporating the L2 norm and the TV norm, which can simultaneously smooth noise information and enhance target edge information. The paper “Q. Zhang, Y. Zhang, Y. Huang, Y. Zhang, J. Pei, Q. Yi, W. Li, and J. Yang, 'Tv-sparse super-resolution method for radar forward-looking imaging,' IEEE Transactions on Geoscience and Remote Sensing, vol. 58, no. 9, pp. 6534-6549, 2020” uses the L1 norm and TV norm as combined constraint terms to enhance the target resolution and edge information. However, as the number of regularization terms increases, the selection of regularization parameters becomes complex.

[0004] In summary, while the methods mentioned above can improve the angular resolution of real aperture radar to some extent, the performance improvement of a single regularization method is limited, and existing combined regularization methods face complex parameter selection during the solution process, which is not conducive to practical applications. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention proposes an adaptive combined regularized angular super-resolution imaging method for real aperture radar to reconstruct the scale information of extended targets. By selecting the generalized sparse norm and the generalized total variation norm as the combined norm, the angular resolution and the extended target edge information are enhanced simultaneously. Then, an iterative reweighting method based on the data correlation fitting criterion is proposed to achieve adaptive and rapid selection of regularization parameters.

[0006] The technical solution of this invention is: an adaptive combined regularized angle super-resolution imaging method for real aperture radar, the specific steps of which are as follows:

[0007] Step 1: Acquisition and preprocessing of echo data;

[0008] Based on the motion geometry model of airborne real aperture radar, the radar acquires information about the target scene Ω by transmitting a linear frequency modulated signal. After demodulation, the echo of the target scene can be expressed as:

[0009]

[0010] Where τ represents the range-direction time sampling vector, t represents the azimuth-direction time sampling vector, the number of azimuth sampling points in the observed scene Ω is N, and the number of range sampling points is M; σ(x,y) represents the target scattering coefficient at point (x,y) in scene Ω, w(t) represents the antenna pattern function modulation, rect(·) represents the rectangular window function, and T p Let λ represent the pulse width of the transmitted signal, λ represent the carrier wavelength, c represent the electromagnetic wave propagation speed, k represent the linear frequency modulation frequency, and n(τ,t) represent additive white Gaussian noise. The target's range history is... R0 represents the initial distance to the target, v represents the speed of the airborne platform, and θ0 represents the spatial azimuth of the target.

[0011] Pulse compression and range travel correction are applied to the echoes to achieve high-resolution range processing. The azimuth echoes of the target scene can be transformed into the following form:

[0012] y = Hx + n (2)

[0013] Where y represents the received azimuth echo vector, H represents the convolution measurement matrix composed of the antenna pattern function, x represents the target scattering coefficient distribution, and n represents the noise vector.

[0014] Step 2: Construct the regularization objective function;

[0015] Within the regularization framework, the generalized sparse norm and the generalized total variation norm are chosen as the combinatorial norms to construct the objective function, which is expressed as follows:

[0016]

[0017] in, This represents the recovered target scattering coefficient distribution. Represents data characteristics. η2 represents the square of the vector's second norm; η1 and η2 represent the regularization parameters; Represents the generalized sparse norm. Let represent the generalized total variation norm, D represent the gradient matrix, and p and q represent the norm values ​​of the generalized sparse norm and the generalized total variation norm, respectively.

[0018] Step 3: Calculate the normalized weighted matrix;

[0019] According to the covariance fitting criterion, the regularization parameters η1 and η2 in equation (3) can be replaced by a weighting matrix, as shown in the following expression:

[0020]

[0021] Where W represents the weighted matrix, and its normalized representation is:

[0022] W = diag(ω1, ω2, ..., ω N (5)

[0023] Where, diag(·) denotes a diagonal matrix, and a n The convolution measurement matrix H = [a1, ..., a2] represents the total convolution measurement matrix. n ,…a N The nth column of the diagram, where N represents the number of azimuth sampling points.

[0024] Step 4: Iterative initialization;

[0025] Initial value of target scattering coefficient for:

[0026]

[0027] Where T represents the transpose of the matrix.

[0028] Step 5: Update the weighting factors;

[0029] W p and W q Let the regularization weight matrices for the p-norm and q-norm be represented respectively, and updated to the following forms:

[0030]

[0031]

[0032] in, Let p-2 be the absolute value of the nth target scattering coefficient element in the (i-1)th iteration; Let q-2 represent the absolute value of the nth target scattering coefficient element in the (i-1)th iteration.

[0033] Step 6: Update the target scattering coefficient;

[0034] Based on the regularization weight matrix in step five, the target scattering coefficients The update and iteration formula is:

[0035]

[0036] Repeat steps five and six for I iterations until the error between two consecutive super-resolution results is no greater than the set error value χ, at which point the loop ends. The loop termination condition is:

[0037]

[0038] By iterating through all range cells of the echo matrix, the azimuth super-resolution imaging results of the entire echo matrix are obtained.

[0039] The beneficial effects of this invention are as follows: The method of this invention first acquires and preprocesses the echo data, constructs a regularized objective function, then calculates the normalized weighted matrix, performs iterative initialization, updates the weighting factors, and finally updates the target scattering coefficients to obtain the azimuth super-resolution imaging result of the entire echo matrix. This invention combines the generalized sparse norm and the generalized total variation norm as constraint terms, simultaneously enhancing angular resolution and expanding target edge information. Then, it employs a data-driven iterative reweighting method, avoiding the selection of regularization parameters and reducing the number of manually selected parameters. Compared with existing combined norm methods, it has a stronger scale information reconstruction capability and improves the scale reconstruction accuracy of real aperture radar for extended targets. Attached Figure Description

[0040] Figure 1 This is a flowchart of an adaptive combined regularized angle super-resolution imaging method for real aperture radar according to the present invention.

[0041] Figure 2 This is a geometric model diagram of the motion of an airborne real aperture radar in an embodiment of the present invention.

[0042] Figure 3 The image shows the original distribution and original echo (SNR = 15dB) of the target in this embodiment of the invention.

[0043] Figure 4 This is a comparison chart of the extended target reconstruction results of various methods in the embodiments of the present invention. Detailed Implementation

[0044] This invention uses simulation experiments to demonstrate the effectiveness of the proposed method. All steps and conclusions of this invention are verified on the Matlab2018 simulation platform.

[0045] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0046] like Figure 1 The flowchart of the adaptive combined regularized angle super-resolution imaging method for real aperture radar of the present invention is shown below. The specific steps are as follows:

[0047] Step 1: Echo data acquisition and preprocessing;

[0048] like Figure 2 As shown in Table 1, this invention employs an airborne real-aperture radar motion geometry model, and the specific parameter values ​​of the airborne platform system are shown in Table 1. To simulate the low signal-to-noise ratio environment of real-world conditions, a noise level of 15 dB was added to this simulation.

[0049] Table 1

[0050] Simulation parameters numerical values carrier frequency 10GHz Time width 2us bandwidth 75MHz Antenna beamwidth 3° Pulse repetition frequency 1000Hz Scan speed 30° / s Scan range ±10° Target distance 5km sampling frequency 100MHz

[0051] In this embodiment, the simulated scanning detection area is set to Ω = -10° to 10°, the antenna beamwidth is 3°, and the scanning speed is 30° / s.

[0052] The initial position of the radar platform is (0,0,H), and the relative position of target E is (R0,θ0,φ0). The range history of this target is R(t), expressed as:

[0053]

[0054] If a radar transmits a linear frequency modulated signal, the received signal can be expressed as:

[0055]

[0056] Where, r∈Ω r and θ∈Ω θ Ω represents the distance variable and the angle variable, respectively. r and Ω θ Let σ(r,θ) represent the range and orientation dimensions of the target scene Ω, respectively. σ(r,θ) represents the target scattering coefficient at (r,θ); τ (r , θ) =2R(t) / c, and c = 3 × 10 8 m / s; h(θ) represents the antenna pattern function; the pulse width T of the transmitted signal p =2μs; carrier frequency f0 = 10GHz, K r This indicates frequency tuning, and The signal bandwidth B = 75MHz.

[0057] After pulse compression and motion compensation processing, the azimuth echo of the target scene can be transformed into the following form:

[0058] y = Hx + n (13)

[0059] Where y represents the received azimuth echo vector, H represents the convolution measurement matrix composed of the antenna pattern function, x represents the target scattering coefficient distribution, and n represents the noise vector.

[0060] Step 2: Construct the regularization objective function;

[0061] Within the regularization framework, this embodiment uses the generalized sparse norm and the generalized total variation norm as a combined norm to construct the objective function, the expression of which is:

[0062]

[0063] in, This represents the recovered target scattering coefficient distribution. Represents data characteristics. η2 represents the square of the vector's second norm; η1 and η2 represent the regularization parameters; Represents the generalized sparse norm. Let represent the generalized total variation norm, D represent the gradient matrix, and p and q represent the norm values ​​of the generalized sparse norm and the generalized total variation norm, respectively. In this embodiment, p = 0.2 and q = 0.8 are selected.

[0064] Step 3: Calculate the normalized weighted matrix;

[0065] According to the covariance fitting criterion, the regularization parameters η1 and η2 in equation (13) are replaced by a weighted matrix, and the specific expression is as follows:

[0066]

[0067] Where W represents the weighted matrix, and its normalized representation is:

[0068] W = diag(ω1, ω2, ..., ω N (16)

[0069] Where, diag(·) denotes a diagonal matrix, and a n The convolution measurement matrix H = [a1, ..., a2] represents the total convolution measurement matrix. n ,…a N The nth column of the diagram, where N represents the number of azimuth sampling points.

[0070] Step 4: Iterative initialization;

[0071] Initial value of the target scattering coefficient of the initial range cell for:

[0072]

[0073] Where T represents the transpose of the matrix.

[0074] Step 5: Update the weighting factors

[0075] W p and W q Let the regularization weight matrices for the p-norm and q-norm be represented respectively, and updated to the following forms:

[0076]

[0077]

[0078] in, Let p-2 be the absolute value of the nth target scattering coefficient element in the (i-1)th iteration; Let q-2 represent the absolute value of the nth target scattering coefficient element in the (i-1)th iteration, where i represents the iteration number.

[0079] Step 6: Update the target scattering coefficient;

[0080] Based on the regularization weight matrix in step five, the target scattering coefficients The update and iteration formula is:

[0081]

[0082] Repeat steps five and six for I iterations until the error between two consecutive super-resolution results is no greater than the set error value χ, at which point the loop ends. The loop termination condition is:

[0083]

[0084] By iterating through all range cells of the echo matrix, the azimuth super-resolution imaging results of the entire echo matrix are obtained.

[0085] like Figure 3 As shown, Figure 3 (a) represents the original scene. Figure 3 (b) represents the actual echo data at SNR = 15 dB. To avoid the randomness of a single reconstruction result, Monte Carlo simulation was used for multiple verifications. Figure 4 This represents the extended target reconstruction results of various methods. Figure 4 (a) shows the reconstruction result using the TV norm regularization method. Figure 4(b) shows the reconstruction results using the combined L2 norm and TV norm regularization method. Figure 4 (c) shows the reconstruction results using the combined L1 norm and TV norm regularization method. Figure 4 (d) shows the reconstruction result of the method of the present invention, where p = 0.2 and q = 0.8; from Figure 4 As can be seen, the reconstruction results of the method of the present invention are superior to other traditional methods in terms of scale fidelity, and the reconstruction results are more robust.

[0086] In summary, the method of this invention improves the scale reconstruction accuracy of real aperture radar for extended targets. First, the azimuth echo is modeled as the convolution of the target scattering distribution and the antenna radiation function. Second, within a regularization framework, a generalized sparse norm is used to enhance target resolution, while a generalized total variation norm is used to ensure the scale of target reconstruction. Finally, a data correlation fitting criterion is used to adaptively select the penalty parameters of the iterative reweighting method, reducing the number of manually selected parameters. The innovation of this invention lies in combining the generalized sparse norm and the generalized total variation norm as constraint terms. Compared with existing combined norm methods, it has a stronger scale information reconstruction capability, and the iterative reweighting method based on the data correlation fitting criterion reduces the number of manually selected parameters.

[0087] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the scope of the claims of the invention.

Claims

1. An adaptive combined regularized angle super-resolution imaging method for real aperture radar, the specific steps of which are as follows: Step 1: Acquisition and preprocessing of echo data; Based on the motion geometry model of airborne real aperture radar, the radar acquires the target scene by transmitting linear frequency modulated signals. After demodulation, the echo of the target scene can be represented as: ； in, Represents the distance-to-time sampling vector. Represents the azimuth time sampling vector, the observation scene. The number of azimuth sampling points is N, and the number of range sampling points is M; Representing a scene middle The target scattering coefficient at a point. This indicates the modulation of the antenna pattern function. Represents a rectangular window function. Indicates the pulse width of the transmitted signal. Indicates the carrier frequency wavelength. Indicates the speed of electromagnetic wave propagation. Indicates linear frequency modulation. This represents additive white Gaussian noise; the target's distance history is... , Indicates the starting distance of the target. Indicates the speed of the airborne platform. Indicates the target's spatial azimuth; Pulse compression and range travel correction are applied to the echoes to achieve high-resolution range processing, transforming the azimuth echoes of the target scene into the following form: ； Where y represents the received azimuth echo vector, and H represents the convolution measurement matrix composed of the antenna pattern function. Indicates the distribution of target scattering coefficients. Represents a noise vector; Step 2: Construct the regularization objective function; Within the regularization framework, the generalized sparse norm and the generalized total variation norm are chosen as the combinatorial norms to construct the objective function, which is expressed as follows: ； in, This represents the recovered target scattering coefficient distribution. Represents data characteristics. This represents the square of the second norm of a vector; and Represents the regularization parameter; Represents the generalized sparse norm. Denotes the generalized total variation norm. Represents the gradient matrix. and Let represent the norm values ​​of the generalized sparse norm and the generalized total variation norm, respectively; Step 3: Calculate the normalized weighted matrix; According to the covariance fitting criterion, the regularization parameter in equation (3) and This can be replaced by a weighted matrix, as shown in the following expression: ； in, The weighted matrix is ​​represented as follows: ； in, Denotes a diagonal matrix, and , Represents the convolution measurement matrix The List, Indicates the number of azimuth sampling points; Step 4: Iterative initialization; Initial value of target scattering coefficient for: ； Where T represents the transpose of the matrix; Step 5: Update the weighting factors; and They represent Norm and The norm-regularized weight matrix is ​​updated to the following forms: ； ； in, Indicates the first In the nth iteration The absolute value of each target scattering coefficient element Power; Indicates the first In the nth iteration The absolute value of each target scattering coefficient element Power; Step 6: Update the target scattering coefficient; Based on the regularization weight matrix in step five, the target scattering coefficients The update and iteration formula is: ； Repeat steps five and six, and then... After several iterations, until the error between two consecutive super-resolution results is no greater than the set error value. The loop ends; the loop termination condition is: ； By iterating through all range cells of the echo matrix, the azimuth super-resolution imaging results of the entire echo matrix are obtained.

Images