MIMO multi-view radar correlation imaging method and device based on image entropy and medium

Abstract

The application discloses a kind of MIMO multi-view radar correlation imaging method, device and medium based on image entropy, for the new problem that radar correlation imaging unit scattering intensity fluctuation with uncertainty of view angle causes radar echo and corresponding imaging unit reference signal correlation mismatch, using the strategy of "result-oriented", from the energy dispersion of imaging unit, based on the sparse reconstruction of imaging scene, join image entropy to construct new optimization model, and using soft threshold iterative shrinkage algorithm to solve the optimization model, obtain the multi-view MIMO radar correlation imaging result of high resolution.The application increases the number of imaging channels, expands the observation angle, increases the randomness of radiation field, and can achieve high-precision imaging under the condition of unit scattering intensity fluctuation.

Description

Technical Field

[0001] This invention belongs to the field of radar correlation imaging signal processing, specifically relating to a MIMO multi-view radar correlation imaging method, device, and medium based on image entropy. Background Technology

[0002] Radar correlation imaging, originating from optical ghost imaging, is a novel type of real-aperture imaging radar. The transmitter emits multiple spatiotemporally uncorrelated random modulated signals, forming a two-dimensional random radiation field within the imaging region. The image is reconstructed by correlating the echo signals with a derived reference signal matrix. Compared to traditional real-aperture radar imaging, its imaging resolution is related to the randomness of the radiation field, eliminating the need for a large antenna array. Furthermore, it is no longer limited by relative rotation with the target, overcoming the limitations of relative-motion imaging radar, which is both dependent on and constrained by motion. Radar correlation imaging provides a new approach and direction for addressing the shortcomings of traditional radar imaging, attracting widespread attention from scholars both domestically and internationally.

[0003] MIMO (Multiple Input Multiple Output) radar correlation imaging improves resolution by increasing the number of receiving channels and expanding the observation angle. However, the fluctuating target scattering coefficient with changing viewing angle can interfere with radar correlation imaging. Existing traditional radar imaging algorithms, as well as radar correlation imaging, are mostly based on isotropic ideal point target models. However, radar correlation imaging heavily relies on the correlation between the radar echo and the reference signal of the corresponding imaging resolution unit. Therefore, in MIMO or MISO (Multiple Input Single Output) radar correlation imaging, fluctuations in the scattering intensity of the resolution unit with viewing angle uncertainty can cause a mismatch in the correlation between the radar echo and the reference signal of the corresponding imaging resolution unit, leading to energy dispersion in the imaging resolution unit and severely affecting the quality of radar correlation imaging.

[0004] Radar images of complex targets are estimates of the target's scattering distribution function obtained through weighted and coherent integration of broadband scattering measurement data. Some scholars have introduced fundamental concepts of target scattering function and scattering distribution function consistent with the traditional definition of Radar Cross Section (RCS), combining classical target scattering mechanisms and radar image analysis to discuss the understanding of high-resolution radar images of complex targets and the interpretation of pixel values. They have pointed out that the pixel value of a radar image resolution unit should not be directly interpreted as the target's RCS level. The scattering intensity of a radar imaging resolution unit is affected by the radar radiation frequency, target material, overall shape, and the resolution unit's position within the overall target, exhibiting uncertain fluctuations with changing viewing angle. The fluctuations in radar image resolution unit scattering intensity with viewing angle differ from the fluctuations in the overall target RCS with viewing angle and cannot be described using a Swerling fluctuation model of the overall target RCS in the time dimension. Therefore, we choose a non-uniform distribution to characterize the fluctuations in radar image resolution unit scattering intensity with changing viewing angle.

[0005] Currently, some scholars have proposed a sparse total variation regularization algorithm to address the impact of fluctuations in the scattering intensity of resolution cells on radar correlation imaging. In the sparse reconstruction model, a total variation term is added to constrain the fluctuation energy. Although this method can suppress the fluctuation energy to a certain extent and improve the visual effect of radar images, the target resolution cells often have local "defocus".

[0006] This paper addresses the shortcomings of existing research by proposing a MIMO multi-view radar correlation imaging algorithm based on image entropy. The algorithm adopts a "results-oriented" strategy, starting with suppressing the energy dispersion of imaging resolution cells. Based on sparse reconstruction of the imaging scene, image entropy is incorporated to construct a new optimization model. An iterative shrinkage-thresholding algorithm (ISTA) is then used to solve this optimization model, yielding high-resolution multi-view MIMO radar correlation imaging results. Simulation results demonstrate that this invention can achieve high-precision imaging even with fluctuations in the scattering intensity of resolution cells. Summary of the Invention

[0007] Purpose of the invention: To address the problem of image quality degradation caused by fluctuations in the scattering intensity of the target region resolution unit in MIMO multi-view radar correlation imaging, a method, device, and medium based on image entropy are proposed for MIMO multi-view radar correlation imaging.

[0008] Technical solution: The MIMO multi-view radar correlation imaging method based on image entropy described in this invention specifically includes the following steps:

[0009] (1) The number of transmitting array elements of the MIMO system radar is M, and the number of receiving array elements is N. A set of spatiotemporally uncorrelated random modulation signals are transmitted at the transmitting end to divide the imaging plane into L imaging units. Based on the positions of the transmitting and receiving array elements and the positions of the imaging units, N reference matrices are obtained. The N reference matrices are spliced ​​in the time dimension to obtain the extended dimension reference matrix A.

[0010] (2) A Gaussian mixture distribution model is used to characterize the fluctuation of the scattering intensity of the radar image resolution unit with the viewing angle; N receiving array elements receive N echo signals, and the N echo signals are spliced ​​in the time dimension to obtain the extended dimension echo signal y;

[0011] (3) Under the premise of target sparsity, a MIMO multi-view radar correlation imaging model based on image entropy is constructed, the iteration number is set to k=0, and the initial target scattering sparse vector σ is set to 0. 0 =0;

[0012] (4) Based on the MIMO multi-view radar correlation imaging model based on image entropy, the extended dimension reference matrix A and the extended dimension echo signal y are jointly processed, and the soft threshold iterative shrinkage algorithm ISTA is used to solve the problem. In the (k+1)th iteration, based on the solution σ of the kth iteration... k The scattering coefficient value σ for multi-view targets is obtained. k+1 The estimate;

[0013] (5) Set the maximum number of iterations K max Given a convergence threshold η, let k = k + 1, and determine whether the number of iterations has been reached or the convergence condition has been met. If satisfied, the comprehensive value of the MIMO multi-view radar associated imaging scattering coefficient is obtained; if not satisfied, return to step (4).

[0014] (6) The obtained target scattering coefficient vector σ is converted back into the scattering coefficient matrix of the two-dimensional imaging region resolution unit to realize MIMO multi-view radar correlation imaging.

[0015] Furthermore, the implementation process of step (1) is as follows:

[0016] The m-th transmitting element transmits signal St. m (t), then the reference signal of the nth receiving element and the mth transmitting element with respect to the lth grid is:

[0017]

[0018] Among them, R m and R n These are the position vectors of the m-th transmitting element and the n-th receiving element, respectively, r lLet c be the position vector of the center of the l-th grid, c be the speed of light, and t be time; let t be the echo signal received by the n-th receiving element from the m-th transmitting element.

[0019]

[0020] Among them, w m,n (t) represents the noise signal of this channel; σ m,n,l The scattering coefficient corresponding to the center of the l-th resolution cell from this viewpoint is discretized over time, with J time samples. The echo signal and reference signal received by the n-th receiving element from the m-th transmitting element are discretized in the time dimension. The MIMO multi-view radar correlation imaging equation is written as:

[0021] y m,n =A m,n σ m,n +w m,n

[0022]

[0023] Among them, y m,n A is the echo vector received by the nth receiving element from the mth transmitting element; m,n This is the reference matrix derived from this perspective; σ m,n =[σ m,n,1 σ m,n,2 …σ m,n,L ] T The scattering coefficient vector for this viewpoint; w m,n This is the noise vector;

[0024] Based on the positions of the transceiver array elements and the imaging unit positions, N reference matrices A1, A2, ..., A1 are obtained. N By concatenating the N reference matrices along the time dimension, we obtain the extended-dimensional reference matrix A = [A1, A2, ..., A...]. N ] T .

[0025] Furthermore, the implementation process of step (2) is as follows:

[0026] The scattering intensity vector σ of a certain resolving element at different viewing angles l =[σ l,1 … σ l,M×N The model follows a Gaussian mixture distribution, and its probability density function is expressed as follows:

[0027]

[0028] Where K represents the number of Gaussian components; μ i ,Σ iπ represents the mean and covariance matrix of each Gaussian component. i The weighting coefficients representing each Gaussian component;

[0029] For the nth receiving element, the received echo signal comes from all M transmitting elements:

[0030]

[0031] Where, σ n The overall imaging result of the nth receiving element; Δσ m,n =σ m,n -σ n This indicates that for the nth receiving channel, the imaging result corresponding to the mth transmitting channel is related to σ. n The difference, i.e., the fluctuation term from that perspective; A n This is the reference matrix corresponding to the nth receiving array element;

[0032] Using a MIMO system, echo data from N receiving channels were obtained. A data-layer dimension-expanding fusion method was employed to stitch the echo data from multiple receiving channels together in the time dimension. By expanding the spatial dimension, the randomness of the radiation field reference signal was improved. The following equation derives the dimension-expanding imaging equation based on the scattering intensity fluctuations of the resolution cell:

[0033]

[0034] remember

[0035]

[0036] Where y is the extended-dimensional echo vector of MIMO radar correlation imaging; A n Let diag(A) be the reference matrix derived from the nth receiving element, where A1, A2, ..., A N A block diagonal matrix with diagonal elements; For A block diagonal matrix with diagonal elements; Δσ n =σ n -σ represents the imaging result σ corresponding to the nth receiving channel. n The difference σ between the integrated inversion imaging result and the MIMO system.

[0037] Furthermore, the implementation process of step (3) is as follows:

[0038] In the correlation imaging model, finding a solution that minimizes the number of non-zero elements in the scattering coefficient matrix allows the objective function to be written as minimizing the l0 norm of σ. By replacing the objective function in the optimization problem with the l1 norm, the model is transformed into:

[0039]

[0040] In the formula, The l1 norm optimization method is a linear convex problem;

[0041] A new definition of image entropy under the sparse reconstruction system is adopted, replacing the probability defined by the frequency of gray values ​​in traditional image entropy with a probability defined by l. p Entropy is defined in norm form; for a radar imaging space consisting of L imaging grids, the scattering coefficient of the l-th grid is σ. l The corresponding probability is get

[0042] By incorporating image entropy into the objective function, we obtain the imaging model of MIMO radar correlation imaging under the fluctuation of target resolution cell scattering intensity:

[0043]

[0044] Choosing λ1 and λ2, where λ2 = μλ1, we can represent the above equation in a more general way to obtain the target expression:

[0045]

[0046] Furthermore, the implementation process of step (4) is as follows:

[0047] set up In the (k+1)th iteration, based on the solution σ of the kth iteration k Using quadratic approximation:

[0048]

[0049] Where t=(2λ max (A H A)) -1 After adding the l1 norm and image entropy constraints, the optimized model is:

[0050]

[0051] make For h(|σ l |) in Perform a first-order Taylor expansion: Based on the principles of ISTA, the final iterative formula is obtained:

[0052]

[0053] The device according to the present invention includes a memory and a processor, wherein:

[0054] Memory is used to store computer programs that can run on a processor;

[0055] The processor is configured to, while running the computer program, execute the steps of the image entropy-based MIMO multi-view radar correlation imaging method as described above.

[0056] The present invention provides a storage medium storing a computer program, which, when executed by at least one processor, implements the steps of the image entropy-based MIMO multi-view radar correlation imaging method as described above.

[0057] Beneficial Effects: Compared with existing technologies, the beneficial effects of this invention are as follows: This invention adopts a MIMO radar correlation imaging system, which, compared with the MISO correlation imaging system, increases the number of imaging channels, expands the observation angle, and increases the randomness of the radiation field, while also providing a more comprehensive representation of complex targets; Starting from suppressing the energy dispersion of imaging resolution units, this invention incorporates image entropy to construct a new optimization model based on sparse reconstruction of the imaging scene, and uses ISTA to solve this optimization model, which can reduce the impact of target resolution unit scattering intensity fluctuations on imaging quality, resulting in high-resolution multi-view MIMO radar correlation imaging results. Attached Figure Description

[0058] Figure 1 This is a flowchart of a MIMO multi-view radar correlation imaging method based on image entropy;

[0059] Figure 2 This is a schematic diagram of the MIMO radar correlation imaging principle of the present invention;

[0060] Figure 3 These are the imaging results of each method when the scattering intensity of the target resolution unit does not fluctuate;

[0061] Figure 4 The images show the imaging results of various methods when the fluctuation of the scattering intensity of the target resolution unit satisfies a Gaussian mixture distribution of mean and variance as the receiving viewing angle changes. Detailed Implementation

[0062] The present invention will now be described in further detail with reference to the accompanying drawings.

[0063] like Figure 1 As shown, this invention proposes a MIMO multi-view radar correlation imaging method based on image entropy. Based on the l1-norm sparse reconstruction model, image entropy is added to construct a new optimized model, and ISTA is used to solve this optimized model to obtain high-resolution multi-view MIMO radar correlation imaging results. The specific implementation process is as follows:

[0064] Step 1: As Figure 2As shown, a MIMO radar has M transmitting elements and N receiving elements. It transmits a set of spatiotemporally uncorrelated random modulated signals, dividing the imaging plane into L imaging units. The reference signals for the nth receiving element and the mth transmitting element with respect to the lth grid can be written as:

[0065]

[0066] Among them, St m (t) represents the signal transmitted by the m-th transmitting element, R m and R n These are the position vectors of the m-th transmitting element and the n-th receiving element, respectively, r l Let be the position vector of the center of the l-th grid, c be the speed of light, and t be time. The echo signal received by the n-th receiving element from the m-th transmitting element is denoted as:

[0067]

[0068] Among them, w m,n (t) represents the noise signal of this channel; σ m,n,l For the scattering coefficient corresponding to the center of the nth resolution cell from this viewpoint, discretize it over time with J time samples. Discretize the echo signal and reference signal received by the nth receiving element from the mth transmitting element in the time dimension. The MIMO multi-view radar correlation imaging equation can be written as:

[0069] y m,n =A m,n σ m,n +w m,n

[0070]

[0071] Among them, y m,n A is the echo vector received by the nth receiving element from the mth transmitting element; m,n This is the reference matrix derived from this perspective; σ m,n =[σ m,n,1 σ m,n,2 …σ m,n,L ] T The scattering coefficient vector for this viewpoint; w m,n This is the noise vector.

[0072] Based on the positions of the transceiver array elements and the imaging unit positions, N reference matrices A1, A2, ..., A1 are obtained. N By concatenating the N reference matrices along the time dimension, we obtain the extended-dimensional reference matrix A = [A1, A2, ..., A...]. N ] T .

[0073] Step 2: Use the Gaussian mixture distribution model to characterize the fluctuation of the scattering intensity of the radar image resolution unit with the viewing angle, and calculate the extended dimension echo signal y.

[0074] Radar images of complex targets are estimates of the target scattering distribution function obtained by weighting and coherently integrating broadband scattering measurement data. Some scholars have introduced basic concepts of target scattering function and scattering distribution function consistent with the traditional RCS definition, combining classical target scattering mechanisms and radar image analysis to discuss the understanding of high-resolution radar images of complex targets and the interpretation of pixel values. Some scholars have pointed out that the pixel value of a radar image resolution unit should not be directly interpreted as the target's RCS level. The scattering intensity of a radar imaging resolution unit is affected by the radar radiation frequency, target material, overall shape, and the resolution unit's position within the overall target, exhibiting uncertain fluctuations with viewing angle. The fluctuation of radar image resolution unit scattering intensity with viewing angle differs from the fluctuation of the overall target RCS with viewing angle and cannot be described by the Swerling fluctuation model of the overall target RCS in the time dimension. The fluctuation of resolution unit scattering intensity, as a non-cooperative error, has strong uncertainty; therefore, a Gaussian mixture distribution model is chosen to characterize the fluctuation of radar image resolution unit scattering intensity with viewing angle. Assume that the scattering intensity vector σ of a resolution unit at different viewing angles... l =[σ l,1 … σ l,M×N The model follows a Gaussian mixture distribution, and its probability density function is expressed as follows:

[0075]

[0076] Where K represents the number of Gaussian components; μ i ,∑ i The mean and covariance matrices representing each Gaussian component are experimentally defined values. For ease of subsequent experiments, let... in π is the diagonal element of the covariance matrix for each Gaussian component; i The weight coefficients for each Gaussian component, and satisfying the following conditions: Gaussian mixture distributions sample based on weights, typically determining which Gaussian component to use as the basis for sampling data based on the weights of each Gaussian component. Gaussian components with larger weights are more likely to be selected, ensuring that samples are more likely to come from Gaussian components with larger weights, thus conforming to the distribution characteristics of a Gaussian mixture distribution model.

[0077] For the nth receiving element, the received echo signal comes from all M transmitting elements:

[0078]

[0079] Where, σ n The overall imaging result of the nth receiving element; Δσ m,n =σ m,n -σ n This indicates that for the nth receiving channel, the imaging result corresponding to the mth transmitting channel is related to σ. n The difference, i.e., the fluctuation term from that perspective; A n This is the reference matrix corresponding to the nth receiving array element.

[0080] Using a MIMO system, echo data from N receiving channels were obtained. A data-layer dimension-expanding fusion method can be used to stitch the echo data from multiple receiving channels together in the time dimension. By expanding the spatial dimension, the randomness of the radiation field reference signal is improved. The following equation derives the dimension-expanding imaging equation based on the scattering intensity fluctuations of the resolution cell:

[0081]

[0082] remember

[0083]

[0084] Where y is the extended-dimensional echo vector of MIMO radar correlation imaging; A n Let diag(A) be the reference matrix derived from the nth receiving element, where A1, A2, ..., A N A block diagonal matrix with diagonal elements; For A block diagonal matrix with diagonal elements; Δσ n =σ n -σ represents the imaging result σ corresponding to the nth receiving channel. n The difference σ between the integrated inversion imaging result and the MIMO system.

[0085] Step 3: Under the premise of target sparsity, construct a MIMO multi-view radar correlation imaging model based on image entropy, set the iteration number k=0, and the initial target scattering sparse vector σ 0 =0.

[0086] According to the law of large numbers, the more samples there are, the more stable the statistical characteristics become. To ensure the quality of the image in relation to fluctuations in the scattering intensity of the resolving unit, it is desirable to control the energy distribution statistically, concentrate the energy in the target resolving unit, and suppress the energy dissipation caused by fluctuations.

[0087] In the correlation imaging model, assuming the target is sparse, the objective function is to minimize the number of non-zero elements in the scattering coefficient matrix among all solutions. This objective function can be expressed as minimizing the l0 norm of σ. A classic algorithm for this model is the matching pursuit algorithm. However, in practical solutions, the l0 norm is not a convex optimization problem, making it difficult to solve. Since the reference signal matrix columns are uncorrelated in correlation imaging, they easily satisfy the RIP criterion. Therefore, we can replace the objective function in the optimization problem with the l1 norm. The model then transforms into:

[0088]

[0089] In the formula The above-mentioned l1 norm optimization method is a linear convex problem that can be solved efficiently using many available algorithms.

[0090] Image entropy, expressed as the bit average of the image's grayscale set, represents the clustering characteristics of the image's grayscale distribution. The smaller the image entropy, the more concentrated the grayscale distribution is on individual grayscale values. Therefore, we aim to control the energy distribution of radar images by minimizing image entropy, concentrating energy on target resolution cells.

[0091] In actual radar imaging, the scattering coefficient of each grid cell is different, unlike grayscale images which have a defined range of grayscale values. If the traditional image entropy definition method is used, regardless of fluctuations, the image entropy obtained for each imaging session is close to logL. This invention adopts a new definition of image entropy under a sparse reconstruction system, replacing the probability defined by the frequency of grayscale values ​​in traditional image entropy with a probability defined by l. p Entropy is defined in norm form. For a radar imaging space consisting of L imaging grids, the scattering coefficient of the l-th grid is σ. l The corresponding probability is get

[0092] By incorporating image entropy into the objective function, we obtain the imaging model of MIMO radar correlation imaging under the fluctuation of target resolution cell scattering intensity:

[0093]

[0094] By choosing appropriate λ1 and λ2 (λ2 = μλ1), the above equation can be characterized in a more general way, yielding the target expression:

[0095]

[0096] Step 4: Based on the MIMO multi-view radar correlation imaging model based on image entropy, the extended dimension reference matrix A and the extended dimension echo signal y are jointly processed. Using the ISTA method, in the (k+1)th iteration, based on the solution σ of the kth iteration... k The mean scattering coefficient σ of the target from multiple viewpoints was obtained. k+1 The estimate.

[0097] This invention employs the Soft Threshold Iterative Shrinkage (ISTA) algorithm to solve the above equations, an algorithm originally proposed for solving the l1 norm model. Let... In the (k+1)th iteration, based on the solution σ of the kth iteration k Using quadratic approximation:

[0098]

[0099] Where t=(2λ max (A H A)) -1 With the addition of the l1 norm and image entropy constraints, the optimization model can be written as:

[0100]

[0101] make For h(|σ l |) in Perform a first-order Taylor expansion:

[0102] make

[0103] When g(σ) l The function is minimized when the partial derivative is 0. When the partial derivative is 0, σ l The value of, that is:

[0104]

[0105] It can be obtained

[0106] when hour, when hour, at this time and With the same modulus, by swapping the independent and dependent variables, we can obtain:

[0107]

[0108] make have:

[0109]

[0110] Step 5: Set the maximum number of iterations K max Given a suitable convergence threshold η (a normal value close to zero, typically less than 10^(-5) in experiments), let k = k + 1, and determine whether the number of iterations has been reached or the convergence condition has been met. If the condition is met, the comprehensive value of the MIMO multi-view radar associated imaging scattering coefficient is obtained; if the condition is not met, return to step 4.

[0111] Step 6: Convert the obtained target scattering coefficient vector σ back into a two-dimensional imaging region resolution cell scattering coefficient matrix using the reshape function in MATLAB, and then display it using the imagesc function.

[0112] The radar images reconstructed by this invention are compared with those reconstructed by traditional methods (correlation method, inverse method, stage singular value decomposition (TSVD), total variation regularization algorithm, orthogonal matching pursuit algorithm (OMP), l1 norm sparse reconstruction method, and sparse total variation radar correlation imaging algorithm). Specific results are as follows: Figure 3 , Figure 4 As shown, when the fluctuations in the scattering intensity of the resolving unit are not considered, all algorithms except the correlation method can achieve relatively accurate reconstruction. When the scattering intensity of the resolving unit fluctuates drastically with the receiving angle, the other algorithms will exhibit varying degrees of "blurring," while the algorithm proposed in this invention can still achieve accurate reconstruction, demonstrating the effectiveness of the algorithm proposed in this invention.

[0113] The present invention also provides an apparatus comprising a memory and a processor, wherein: the memory is used to store a computer program capable of running on the processor; and the processor is used to execute, when running the computer program, the steps of the image entropy-based MIMO multi-view radar correlation imaging method as described above.

[0114] The present invention also provides a storage medium storing a computer program, which, when executed by at least one processor, implements the steps of the image entropy-based MIMO multi-view radar correlation imaging method as described above.

[0115] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A MIMO multi-view radar correlation imaging method based on image entropy, characterized in that, Includes the following steps: (1) The number of transmitting array elements of the MIMO system radar is M, and the number of receiving array elements is N. A set of spatiotemporally uncorrelated random modulation signals are transmitted at the transmitting end to divide the imaging plane into L imaging units. Based on the positions of the transmitting and receiving array elements and the positions of the imaging units, N reference matrices are obtained. The N reference matrices are spliced ​​in the time dimension to obtain the extended dimension reference matrix A. (2) A Gaussian mixture distribution model is used to characterize the fluctuation of the scattering intensity of the radar image resolution unit with the viewing angle; N receiving array elements receive N echo signals, and the N echo signals are spliced ​​in the time dimension to obtain the extended dimension echo signal y; (3) Under the premise of target sparseness, a MIMO multi-view radar correlation imaging model based on image entropy is constructed, the iteration number is set to k=0, and the initial target scattering sparse vector σ is set to 0. 0 =0; (4) Based on the MIMO multi-view radar correlation imaging model based on image entropy, the extended dimension reference matrix A and the extended dimension echo signal y are jointly processed, and the soft threshold iterative shrinkage algorithm ISTA is used to solve the problem. In the (k+1)th iteration, based on the solution σ of the kth iteration... k The scattering coefficient value σ for multi-view targets is obtained. k+1 The estimate; (5) Set the maximum number of iterations K max Given a convergence threshold η, let k = k + 1, and determine whether the number of iterations has been reached or the convergence condition has been met. If satisfied, the comprehensive value of the MIMO multi-view radar associated imaging scattering coefficient is obtained; if not satisfied, return to step (4). (6) The obtained target scattering coefficient vector σ is converted back into the scattering coefficient matrix of the two-dimensional imaging region resolution unit to realize MIMO multi-view radar correlation imaging.

2. The MIMO multi-view radar correlation imaging method based on image entropy according to claim 1, characterized in that, The implementation process of step (1) is as follows: The m-th transmitting element transmits signal St. m (t), then the reference signal of the nth receiving element and the mth transmitting element with respect to the lth grid is: Among them, R m and R n These are the position vectors of the m-th transmitting element and the n-th receiving element, respectively, r l Let c be the position vector of the center of the l-th grid, c be the speed of light, and t be time; let t be the echo signal received by the n-th receiving element from the m-th transmitting element. Among them, w m,n (t) represents the noise signal of the receiving array element; σ m,n,l The scattering coefficient corresponding to the center of the l-th resolution cell from this viewpoint is discretized over time, with J time samples. The echo signal and reference signal received by the n-th receiving element from the m-th transmitting element are discretized in the time dimension. The MIMO multi-view radar correlation imaging equation is written as: y m,n =A m,n s m,n +w m,n Among them, y m,n A is the echo vector received by the nth receiving element from the mth transmitting element; m,n This is the reference matrix derived from this perspective; σ m,n =[σ m,n,1 σ m,n,2 … σ m,n,L ] T The scattering coefficient vector for this viewpoint; w m,n This is the noise vector; Based on the positions of the transceiver array elements and the imaging unit positions, N reference matrices A1, A2, ..., A1 are obtained. N By concatenating the N reference matrices along the time dimension, we obtain the extended-dimensional reference matrix A = [A1, A2, ..., A...]. N ] T .

3. The MIMO multi-view radar correlation imaging method based on image entropy according to claim 2, characterized in that, The implementation process of step (2) is as follows: The scattering intensity vector σ of a certain resolving element at different viewing angles l =[σ l,1 … σ l,M×N The model follows a Gaussian mixture distribution, and its probability density function is expressed as follows: Where K represents the number of Gaussian components; μ i ,∑ i π represents the mean and covariance matrix of each Gaussian component. i The weighting coefficients representing each Gaussian component; For the nth receiving element, the received echo signal comes from all M transmitting elements: Where, σ n This represents the overall imaging result of the nth receiving element; Δσ m,n =σ m,n -σ n This indicates that for the nth receiving channel, the imaging result corresponding to the mth transmitting channel is related to σ. n The difference, i.e., the fluctuation term from that perspective; A n This is the reference matrix corresponding to the nth receiving array element; Using a MIMO system, echo data from N receiving channels were obtained. A data-layer dimension-expanding fusion method was employed to stitch the echo data from multiple receiving channels together in the time dimension. By expanding the spatial dimension, the randomness of the radiation field reference signal was improved. The following equation derives the dimension-expanding imaging equation based on the scattering intensity fluctuations of the resolution cell: remember Where y is the extended-dimensional echo vector of MIMO radar correlation imaging; A n Let diag(A) be the reference matrix derived from the nth receiving element, where A1, A2, ..., A N A block diagonal matrix with diagonal elements; For A block diagonal matrix with diagonal elements; Δσ n =σ n -σ represents the imaging result σ corresponding to the nth receiving channel. n The difference σ between the integrated inversion imaging result and the MIMO system.

4. The MIMO multi-view radar correlation imaging method based on image entropy according to claim 3, characterized in that, The implementation process of step (3) is as follows: In the correlation imaging model, finding a solution that minimizes the number of non-zero elements in the scattering coefficient matrix allows the objective function to be written as minimizing the l0 norm of σ. By replacing the objective function in the optimization problem with the l1 norm, the model is transformed into: In the formula, The l1 norm optimization method is a linear convex problem; A new definition of image entropy under the sparse reconstruction system is adopted, replacing the probability defined by the frequency of gray values ​​in traditional image entropy with a probability defined by l. p Entropy is defined in norm form; for a radar imaging space consisting of L imaging grids, the scattering coefficient of the l-th grid is σ. l The corresponding probability is get By incorporating image entropy into the objective function, we obtain the imaging model of MIMO radar correlation imaging under the fluctuation of target resolution cell scattering intensity: Choosing λ1 and λ2, where λ2 = μλ1, we can represent the above equation in a more general way to obtain the target expression:

5. The MIMO multi-view radar correlation imaging method based on image entropy according to claim 4, characterized in that, The implementation process of step (4) is as follows: set up In the (k+1)th iteration, based on the solution σ of the kth iteration k Using quadratic approximation: Where t=(2λ max (A H A)) -1 After adding the l1 norm and image entropy constraints, the optimized model is: make For h(|σ l |) in Perform a first-order Taylor expansion: Based on the principles of ISTA, the final iterative formula is obtained:

6. A device, characterized in that, Includes memory and processor, wherein: Memory is used to store computer programs that can run on a processor; A processor, configured to, while running the computer program, perform the steps of the image entropy-based MIMO multi-view radar correlation imaging method as described in any one of claims 1 to 5.

7. A storage medium, characterized in that, The storage medium stores a computer program that, when executed by at least one processor, implements the steps of the MIMO multi-view radar correlation imaging method based on image entropy as described in any one of claims 1 to 5.

Images