Online adaptive radar rapid imaging method

Abstract

The invention provides an online adaptive radar rapid imaging method, and relates to the technical field of radar imaging, and the method comprises the steps: building an azimuth echo convolution model, and carrying out the restoration processing of an antenna measurement matrix, so as to reduce the imaging edge information error; recursive updating of local beams is realized based on independence of echo data in adjacent beam widths and structural characteristics of a repaired antenna measurement matrix; global beam recursion updating is realized through overlapping shift processing; and based on the optimal generalized ridge estimation theory, adaptively selecting regularization parameters, and completing target scene reconstruction. Compared with a traditional angle super-resolution method, the method has the advantages that the corresponding relation between the echo sample and the target grid is utilized, the inherent advantages of an online frame can be kept, and meanwhile the problem that the calculation redundancy of a traditional online method is too large is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of radar imaging technology, and more specifically to an online adaptive radar rapid imaging method. Background Technology

[0002] Real aperture radar can quickly acquire all-around target detection information through radar beam scanning, but its angular resolution is limited by the antenna size and is usually low.

[0003] To improve the angular resolution of real aperture radar, X.Yin et al. proposed an online reconstruction method based on sparse regularization. This method realizes the online structure of the batch regularization method and achieves real-time imaging. However, since it does not consider the correspondence between different echo samples and target grids, the iterative relationship remains fixed, resulting in a large amount of computational redundancy. Summary of the Invention

[0004] To address the technical problems in related technologies, this invention provides an online adaptive radar fast imaging method to solve the problem that the online reconstruction method based on sparse regularization described in the background technology does not consider the correspondence between different echo samples and target grids, thus leading to a large amount of computational redundancy.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An online adaptive radar fast imaging method includes the following steps: Step S1: Establish an azimuth echo convolution model and repair the antenna measurement matrix to reduce imaging edge information errors; Step S2: Based on the independence of echo data within adjacent beamwidths and the structural characteristics of the repaired antenna measurement matrix, recursive updates of local beams are achieved. Step S3: Achieve global beam recursive update through overlap shifting processing; Step S4: Adaptively select regularization parameters based on the optimal generalized ridge estimation theory to complete the reconstruction of the target scene.

[0006] Optionally, establishing the azimuth echo convolution model specifically includes: The first The echo data of each distance cell is modeled as follows: In the formula, Represents echo data, The target scattering coefficient, This represents zero-mean Gaussian white noise. Let represent the convolution matrix composed of antenna pattern samples, and . in, This is a sample of the antenna radiation pattern. The number of samples in the antenna pattern is denoted as , and ,in, For beamwidth, The pulse repetition frequency, This represents the antenna scanning speed.

[0007] Optionally, in step S1, the repair process for the antenna measurement matrix specifically includes: convolution matrix Repair to : Based on the repair Process it.

[0008] Optionally, in step S2, the recursive update of the local beam specifically includes: The full echo region is divided into several unit echo regions; When the When each echo pulse sample arrives, for Perform a windowing operation to extract non-zero elements, where... It is a convolution matrix After being divided into rows, the first OK; The corresponding parameter matrix is ​​expanded along its diagonal, with the expanded elements having a value of zero.

[0009] Optionally, the dimensional change of the target vector update within a unit echo region satisfies: Each online update iterates through an echo sampling pulse, increasing the dimension of the target vector by one order, and updating within each unit echo region. After [number] data points, the dimension of the target vector increases to [number]. Rank.

[0010] Optionally, if the last unit echo region contains less than The data points were also traversed directly.

[0011] Optionally, in step S3, the global beam recursive update specifically includes: The first The target grid for recovering the unit echo region. Divide into 3 equal-length subvectors: ,in, , and If they are subvectors of equal length, then the first... The target grid for recovering a unit echo region is: Then we have: ,in, This indicates overlapping displacement processing. express target vector and The target result superimposed after the target vector is shifted and translated by point L; The global recursive update result for the complete echo region is: In the formula, , This represents the number of azimuth sampling points for the target scene. The number of samples in the antenna pattern. This is the floor function. This represents the summation operation for overlapping displacement processing.

[0012] Optionally, the final target reconstruction result By truncation The length of the middle is It is obtained by using a vector.

[0013] Optionally, in step S4, the regularization parameter The adaptive selection calculation specifically includes: Calculate the regularization parameters of the current local target mesh cell based on the previous local target mesh. : In the formula, This indicates the number of samples in the antenna radiation pattern. Indicates the first The target scattering coefficient value of a local target grid cell. express The The average value of the target scattering coefficient of each local target grid cell.

[0014] Optionally, the regularization parameters of the first edge local target mesh cell are obtained through iterative initialization using the Tikhonov regularization method.

[0015] Beneficial effects: 1. Through the above technical solution, firstly, the method of the present invention can effectively improve the accuracy of imaging edges. Specifically, by repairing the boundary structure of the antenna measurement matrix, the present invention can effectively eliminate the virtual target error caused by insufficient edge echo samples in traditional methods, thereby ensuring the reconstruction quality of edge targets in the imaging area and avoiding information distortion.

[0016] Second, the method of this invention can significantly reduce computational complexity. Specifically, this invention utilizes the independence of adjacent beam data, recursively updating only locally non-zero elements, thereby avoiding the repeated inversion of high-dimensional matrices in traditional online methods and reducing the exponential computational burden.

[0017] Third, the method of this invention can achieve rapid reconstruction of the entire scene. Specifically, this invention decomposes global reconstruction into unit region updates and shift overlays, avoiding overall matrix operations, thereby meeting the real-time processing requirements of radar echo data while maintaining imaging resolution.

[0018] Fourth, the method of this invention can adaptively optimize imaging quality. Specifically, this invention dynamically adjusts the regularization parameter based on the local target scattering characteristics, avoiding manual parameter tuning, thereby enabling stable target reconstruction even under low signal-to-noise ratio conditions and improving imaging robustness.

[0019] 2. Other beneficial effects or advantages of the present invention will be described in detail in the specific embodiments. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] in: Figure 1 This is a flowchart illustrating the steps of an exemplary embodiment of the online adaptive radar fast imaging method provided by the present invention; Figure 2 This is a geometric model of an airborne real aperture radar provided in an exemplary embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the correspondence between echo pulse samples and target grids provided in an exemplary embodiment of the present invention; Figure 4 This is a schematic diagram of the algorithm flow provided by an exemplary embodiment of the present invention; Figures 5 to 12 This is a comparison chart of the results of different imaging methods provided by an exemplary embodiment of the present invention and the method of the present invention, wherein... Figure 5 Original scene Figure 6 For real beam echo, Figure 7 For the TSVD method, Figure 8 For Lucy's method, Figure 9 For the IAA method, Figure 10 For batch processing method, Figure 11 For online method, Figure 12 This is the method of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0023] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0024] Furthermore, the terms "comprising" and "having," and any variations thereof, used in the description of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices. It should also be noted that in embodiments of this invention, the words "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in embodiments of this invention should not be construed as preferred or advantageous over other embodiments or designs. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0025] To facilitate a clearer and more accurate understanding of the technical solution of the present invention by those skilled in the art, the existing related technologies and their technical problems will be further explained below.

[0026] Real aperture radar can quickly acquire all-around target detection information through radar beam scanning, but its angular resolution is limited by the antenna size and is usually low.

[0027] To improve the angular resolution of real aperture radar, the paper "Y.Zha, Y.Huang, Z.Sun, Y.Wang, and J. Yang, “Bayesian deconvolution for angular super-resolution in forward-looking scanning radar,” Sensors, vol.15, no.3, pp.6924–6946, 2015.” proposes a super-resolution method based on a sparse Bayesian framework. This method suppresses the influence of noise through statistical prior distribution. However, this method suffers from target reconstruction failure under low signal-to-noise ratio conditions.

[0028] Subsequently, the paper "Y. Zhang, W. Li, Y. Zhang, Y. Huang, and J. Yang, "A fast iterative adaptive approach for scanning radar angular superresolution," IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 8, no. 11, pp. 5336–5345, 2015" introduced the adaptive method of spectral estimation in array signal processing into RAR imaging. By utilizing the block triangular structure of the echo autocorrelation matrix, it reduced computational complexity while maintaining super-resolution performance. However, this method requires multiple iterative calculations on the high-dimensional matrix, which limits the real-time imaging capability.

[0029] Furthermore, the paper "X. Yin, Y. Huang, Y. Zhang, Y. Zhang, X. Tuo, Y. Zhang, and J. Yang, “Online sparse super-resolution method for radar forward-looking imaging using majorize-minimization,” in IGARSS 2023-2023 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2023, pp. 7129–7132” proposes an online reconstruction method based on sparse regularization. This method realizes the online structure of the batch regularization method and achieves real-time imaging. However, since it does not consider the correspondence between different echo samples and target grids, the iterative relationship remains fixed, resulting in a large amount of computational redundancy.

[0030] To address the shortcomings of the aforementioned online reconstruction methods based on sparse regularization, which fail to consider the correspondence between different echo samples and the target grid, resulting in significant computational redundancy, this invention proposes an online adaptive radar fast imaging method.

[0031] The technical concept of the online adaptive radar fast imaging method of the present invention is as follows: First, the antenna measurement matrix is ​​repaired to reduce imaging edge information errors; then, based on the characteristic that the echo data within the two beamwidths are independent, the structural characteristics of the corrected steering matrix are used to realize the recursive update of the local beam; finally, the global beam is recursively updated through overlap shifting processing, and the target is effectively reconstructed based on the optimal generalized ridge estimation theory.

[0032] Compared with traditional angular super-resolution methods, the method of this invention can achieve real-time imaging while maintaining the advantages of online frames and significantly reducing storage requirements.

[0033] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0034] like Figures 1 to 12 As shown, the present invention provides an online adaptive radar fast imaging method, comprising the following steps: Step S1: Establish an azimuth echo convolution model and repair the antenna measurement matrix to reduce imaging edge information errors; Step S2: Based on the independence of echo data within adjacent beamwidths and the structural characteristics of the repaired antenna measurement matrix, recursive updates of local beams are achieved. Step S3: Achieve global beam recursive update through overlap shifting processing; Step S4: Adaptively select regularization parameters based on the optimal generalized ridge estimation theory to complete the reconstruction of the target scene.

[0035] Through the above technical solution, firstly, the method of the present invention can effectively improve the accuracy of imaging edges. Specifically, by repairing the boundary structure of the antenna measurement matrix, the present invention can effectively eliminate the virtual target error caused by insufficient edge echo samples in traditional methods, thereby ensuring the reconstruction quality of edge targets in the imaging area and avoiding information distortion.

[0036] Second, the method of this invention can significantly reduce computational complexity. Specifically, this invention utilizes the independence of adjacent beam data, recursively updating only locally non-zero elements, thereby avoiding the repeated inversion of high-dimensional matrices in traditional online methods and reducing the exponential computational burden.

[0037] Third, the method of this invention can achieve rapid reconstruction of the entire scene. Specifically, this invention decomposes global reconstruction into unit region updates and shift overlays, avoiding overall matrix operations, thereby meeting the real-time processing requirements of radar echo data while maintaining imaging resolution.

[0038] Fourth, the method of this invention can adaptively optimize imaging quality. Specifically, this invention dynamically adjusts the regularization parameter based on the local target scattering characteristics, avoiding manual parameter tuning, thereby enabling stable target reconstruction even under low signal-to-noise ratio conditions and improving imaging robustness.

[0039] The technical solution of the present invention will be described below with reference to a specific embodiment.

[0040] Step 1: Establishing the Azimuth Echo Convolution Model Assume the airborne radar platform travels at a constant speed Along The radar antenna flies at a constant speed along the axial direction, with an angular velocity of... Scanning the oblique forward-looking area. The platform's flight altitude is... Platform and target The initial slope distance is .exist At any given moment, the distance between the platform and the goal is historical. Represented as: Equation (1) in, and These represent the azimuth and elevation angles, respectively.

[0041] Radar acquires the raw echo of a target by transmitting a linear frequency modulated signal with a large time-bandwidth product. The acquired raw echo undergoes range-dimensional pulse compression and range-travel correction, and the echo signal can be expressed as... Equation (2) in, This indicates the received echo signal. It is time in the azimuth dimension determined by the antenna scanning speed. It is time in the distance dimension related to the speed of light. and These represent the number of sampling points for distance and direction, respectively. It is the first The magnitude of each target, Represents the antenna pattern function. It's the speed of light. It is the carrier frequency. It is a time delay and .

[0042] Based on the scanning imaging process, the echo signal along the azimuth direction can be described as the convolution of the antenna pattern function and the target scattering coefficient, in the... The echo data of each distance cell can be written in convolutional form as follows: Equation (3) In the formula, Represents echo data, The target scattering coefficient, This represents zero-mean Gaussian white noise. The convolution matrix composed of antenna pattern samples can be written as: in, This is a sample of the antenna radiation pattern. The number of samples and ,in, For beamwidth, The pulse repetition frequency, This represents the antenna scanning speed.

[0043] Step 2: Establishment and Transformation of the Objective Function Batch processing sparsity Since the relationship between the echo and the target is as shown in equation (3) in step one, the target reconstruction problem can be regarded as an optimal estimation problem due to the regularization method. Equation (4) However, this is an ill-posed problem, and a direct solution will inevitably lead to noise amplification. Therefore, in order to suppress noise amplification and improve angular resolution, a regularization framework is introduced... Norm as a constraint term: Equation (5) in, It is a regularization parameter that controls the weights added to the constraint terms.

[0044] Since the second term in equation (5) can be equivalently expressed by the minimax criterion as: Equation (6) in, Represents a diagonally loaded matrix. The first goal The result of the next iteration.

[0045] Therefore, equation (5) can be further expressed as Equation (7) And it can be easily solved through iteration: Equation (8) To achieve real-time imaging, we further derive the online structure of the above equation. First, for the convolution matrix... If it is divided into blocks and rows, then ,in yes The Okay. Based on equation (5), in order to achieve online sparse estimation, equation (7) can be modeled equivalently: Equation (9) in, It's an echo. The The nth element, that is, the nth element received in the orientation dimension. Each echo value.

[0046] Based on the above analysis, equation (9) can be equivalently expressed as: Equation (10) in, It is a diagonal matrix. The estimated first The real-time target scattering coefficient at the arrival of each echo, (10) is given by the closed-form solution of the following equation: Equation (11) For the solution in equation (11), let , ,but Therefore, equation (11) can be written as: Equation (12) when At that time, the above parameters can be initialized as follows: Equation (13) when When the time is right, the following formula is used for recursive calculation: Equation (14) Based on the above derivation, batch processing sparsity is obtained. Online structure of regularization algorithms. Compared with traditional batch processing algorithms, online sparse... Regularization algorithms fully utilize the similarity between adjacent echo signals. However, as the dimension of echo data continues to increase, the computational burden grows exponentially due to the need for repeated matrix inversions. When faced with high-dimensional echo matrices, the computation time will exceed the data collection time, making it difficult to achieve real-time target reconstruction and thus completely negating the advantages of the online update framework.

[0047] Step 3: Measurement Matrix Repair Processing To overcome the above problems and achieve rapid online target reconstruction, the measurement matrix first needs to be repaired.

[0048] According to the physical process of scanning and imaging in a RAR system, as follows: Figure 3 As shown, the reconstruction of the target mesh is based solely on its neighboring meshes. Each echo pulse sample determines that a single echo pulse sample can only affect its corresponding adjacent pulses. One target grid, therefore Each echo pulse sample can actually correspond to only one Reconstruction of target grid points.

[0049] Since the number of target grid points recovered from the edge echo pulse samples exceeds the number of echo samples, the antenna measurement matrix is ​​adjusted to reduce the reconstruction error of the edge target grid. Repair to , can be represented as Equation (15) Based on the repaired antenna measurement matrix, the local beam recursive update framework first restores the target network containing virtual edge targets, then the global translation framework updates the full scene image pairs using overlapping shift operations, and finally truncates the real targets to achieve real-time and fast target reconstruction.

[0050] Step 4: Online unit region update based on a local recursive framework Based on the above analysis Each echo pulse sample corresponds to Reconstruction of target grid points. Among them, the middle one... Each target grid point is obtained by traversing its corresponding complete echo. Therefore, the entire echo region along a distance cell... It can be divided into several unit echo regions Equation (16) in, , Represents the floor function, vector The lengths are all Therefore, by reconstructing the target for each unit echo region, target reconstruction for the entire complete echo region can be achieved.

[0051] In equation (11), and The update is a continuous accumulation process, but with each incoming echo pulse, by observing the structure of the repaired antenna measurement matrix in equation (15), it can be seen that... It contains many zero elements, but the non-zero elements are the main part responsible for reconstructing the target.

[0052] Therefore, when the first When each echo pulse sample arrives, for Perform windowing operations to extract the length of... Non-zero elements are represented as Equation (17) Thus, after windowing operations, we can further obtain... , , and However, due to the repaired antenna measurement matrix in equation (15) It is a cyclic shift matrix. and The valid non-zero elements are not aligned in position. This will further lead to and and and The effective information is added in a misaligned manner during the summation process, so the windowed variables obtained above cannot be directly substituted into equation (11) for calculation.

[0053] To maintain mathematical consistency between the target reconstruction process of the windowed recursive variable and the original online update process, when the... When the echo pulse sample arrives Expand along the diagonal of the matrix, with each expanded element having a value of 0, meaning the dimension is 1. of Expanding to dimensions of Similarly, the dimension is of Expand the matrix diagonally to a dimension of of ; dimension is of Expanded to dimensions of ; dimension is of Expanded to dimensions of Its mathematical expression is as follows: Equation (18) In the formula, The dimension is The zero matrix, The dimension is The zero matrix, The dimension is The zero matrix, The dimension is The zero matrix, The dimension is The zero matrix, The dimension is The zero matrix; Based on the above explanation, (14) can be updated as follows: Equation (19) Therefore, with each online update traversing one echo sampling pulse, the target The dimension is increased by one order, and the update is performed within each unit echo region. After 1 data point The dimension will be from the initial The order increases to The order. It is worth noting that if the last unit echo region contains less than... Each data point is traversed directly using the same principle as described above.

[0054] Step 5: Full-scene image reconstruction based on global translation update For the recovery from each unit echo region A target grid, the edges of which Each target grid is also affected by the echo regions of other units.

[0055] Let's assume the first Each unit echo region is denoted as having a dimension of . of Therefore, it can be divided into ,in, , and It is of equal length The vector.

[0056] Similarly, from the first The target grid for recovering a unit echo region is: .Will Defined as overlapping displacement processing, it can be expressed as: Equation (20) Therefore, the global recursive update result for the complete echo region is: Equation (21) in, , This represents the floor function. This represents the summation operation for overlapping shift processing. Simultaneously, due to the repair of the antenna measurement matrix, a virtual target of length *l* exists at the edge; therefore, the final target reconstruction result... By truncation The length of the middle is It is obtained from a vector.

[0057] Step Six: Adaptive Selection of Regularization Parameters The above framework significantly reduces computational complexity and enables rapid imaging within an online framework. However, the imaging quality often depends heavily on the selection of the regularization parameter in equation (5). To avoid complex manual adjustment, an adaptive parameter selection method is proposed based on ridge estimation theory.

[0058] In classical statistics, In the context of regularization, the optimal generalized ridge parameter is calculated as follows: Equation (22) in, The variance can be estimated using the following formula: Equation (23) in, express The first in Each element value Let represent the mean. Based on the above formula, the regularization parameter corresponding to the optimal generalized ridge parameter can be obtained as: Equation (24) Thus, the regularization parameter of each local target mesh cell can be derived by mapping the target scattering coefficient value of the previous local target mesh cell, thereby achieving adaptive selection of the regularization parameter. The initial regularization value of the first edge local target mesh cell is iteratively initialized using the classic Tikhonov regularization method.

[0059] The innovations of this invention are primarily: firstly, the antenna measurement matrix is ​​repaired, reducing imaging edge information errors; secondly, based on the independent characteristics of echo data within the two beamwidths, the structural properties of the corrected steering matrix are used to achieve recursive updates of the local beams; finally, global beam recursive updates are achieved through overlap-shift processing, and based on the optimal generalized ridge estimation theory, effective target reconstruction with adaptive parameters is realized. Compared with traditional angular super-resolution methods, this invention effectively utilizes the correspondence between echo samples and the target grid, maintaining the inherent advantages of online frames while solving the problem of excessive computational redundancy in traditional online methods.

[0060] An exemplary embodiment will be described below with reference to the accompanying drawings.

[0061] Step 1: Establish the azimuth echo convolution model.

[0062] Adopting such Figure 2 The airborne radar forward-looking imaging geometric model shown below uses the radar simulation system parameters listed in Table 1.

[0063] Table 1 Radar Simulation System Parameters The signal-to-noise ratio is set to Assume the airborne radar platform travels at a constant speed. Along The radar antenna flies at a constant speed along the axial direction, with an angular velocity of... Scanning the oblique forward-looking area. The platform's flight altitude is... The initial slant distance between the platform and the target is .exist At any given moment, the distance between the platform and the goal is historical. Represented as: Equation (25) in, and These represent the azimuth and elevation angles, respectively.

[0064] The radar acquires the raw echo of the target by transmitting a linear frequency modulated signal with a large time-bandwidth product. The acquired raw echo undergoes range-dimensional pulse compression and range-travel correction, and the echo signal is represented as... Equation (26) in, This indicates the received echo signal. It is time in the azimuth dimension determined by the antenna scanning speed. It is time in the distance dimension related to the speed of light. and These represent the number of sampling points for distance and direction, respectively. It is the first The magnitude of each target, Represents the antenna pattern function. It's the speed of light. It is the carrier frequency. It is a time delay and .

[0065] Based on the scanning imaging process, the echo signal along the azimuth direction can be described as the convolution of the antenna pattern function and the target scattering coefficient, in the... The echo data of each distance cell can be written in convolutional form as follows: Equation (27) in, Represents echo data, The target scattering coefficient, This represents zero-mean Gaussian white noise. The convolution matrix composed of antenna pattern samples can be written as: in, For the antenna pattern samples, the number of samples From the pulse repetition frequency ( ), beamwidth ( ) and antenna scanning speed Decision, that is .

[0066] Step 2: Establishment and Transformation of the Objective Function Batch processing sparsity Since the relationship between the echo and the target is as shown in Equation (24) above, the target reconstruction problem can be regarded as an optimal estimation problem by regularization. Equation (28) However, this is an ill-posed problem, and a direct solution will inevitably lead to noise amplification. Therefore, in order to suppress noise amplification and improve angular resolution, a regularization framework is introduced... Norm as a constraint term Equation (29) in, It is a regularization parameter that controls the weights added to the constraint terms.

[0067] Since the second term in equation (26) can be equivalently expressed as Equation (30) in, Represents a diagonally loaded matrix. The first goal The result of the next iteration.

[0068] Therefore, equation (26) can be further expressed as Equation (31) And it can be easily solved using the following iterative formula. Equation (32) To achieve real-time imaging, we further derive the online structure of the above equation. First, for the convolution matrix... If it is divided into blocks and rows, then ,in yes The Okay. Based on equation (26), in order to achieve online sparse estimation, equation (28) can be modeled as equivalent: Equation (33) in, It's an echo. The The nth element, that is, the nth element received in the orientation dimension. Each echo value.

[0069] Based on the above analysis, equation (30) can be equivalently expressed as: Equation (34) in, It is a diagonal matrix. The estimated first The real-time target scattering coefficient when the echo arrives is given by the closed-form solution of equation (31) as follows: Equation (35) For the solution in equation (32), let , ,but Therefore, (32) can be written as: Equation (36) when When this happens, the above parameters can be initialized to... Equation (37) when When this happens, recursive calculation can be performed using the following formula. Equation (38) Based on the above derivation, batch processing sparsity is obtained. Online structure of regularization algorithms. Compared with traditional batch processing algorithms, online sparse... Regularization algorithms fully utilize the similarity between adjacent echo signals. However, as the dimension of echo data continues to increase, the computational burden grows exponentially due to the need for repeated matrix inversions. When faced with high-dimensional echo matrices, the computation time will exceed the data collection time, making it difficult to achieve real-time target reconstruction and thus completely negating the advantages of the online update framework.

[0070] Step 3: Measurement Matrix Repair Processing To overcome the above problems and achieve rapid online target reconstruction, the measurement matrix is ​​first repaired.

[0071] According to the physical process of scanning and imaging in a RAR system, as follows: Figure 3 As shown, the reconstruction of the target mesh is based solely on its neighboring meshes. Each echo pulse sample determines that a single echo pulse sample can only affect its corresponding adjacent pulses. One target grid, therefore Each echo pulse sample can actually correspond to only one Reconstruction of target grid points.

[0072] Since the number of target grid points recovered from the edge echo pulse samples exceeds the number of echo samples, the antenna measurement matrix is ​​adjusted to reduce the reconstruction error of the edge target grid. Repair to , can be represented as Equation (39) Based on the repaired antenna measurement matrix, the local beam recursive update framework first restores the target network containing virtual edge targets, then the global translation framework updates the full scene image pairs using overlapping shift operations, and finally truncates the real targets to achieve real-time and fast target reconstruction.

[0073] Step 4: Online unit region update based on a local recursive framework Based on the above analysis Each echo pulse sample corresponds to Reconstruction of target grid points. Among them, the middle one... Each target grid point is obtained by traversing its corresponding complete echo. Therefore, the entire echo region along a distance cell... It can be divided into several unit echo regions: Equation (40) in, , Represents the floor function, vector The lengths are all Therefore, by completing target reconstruction within each unit echo region, target reconstruction for the entire complete echo region can be achieved.

[0074] In equation (32), and The update is a continuous accumulation process, but with each incoming echo pulse, by observing the structure of the repaired antenna measurement matrix in equation (36), it can be seen that... It contains many zero elements, but the non-zero elements are the main part responsible for reconstructing the target.

[0075] Therefore, when the first When each echo pulse sample arrives, for Perform windowing operations to extract the length of... Non-zero elements are represented as Equation (41) Thus, after windowing operations, we can further obtain... , , and However, due to the repaired antenna measurement matrix in equation (15) It is a cyclic shift matrix. and The valid non-zero elements are not aligned in position. This will further lead to and and and The effective information is added in a misaligned manner during the summation process, so the windowed variables obtained above cannot be directly substituted into equation (32) for calculation.

[0076] To maintain mathematical consistency between the target reconstruction process of the windowed recursive variable and the original online update process, when the... When the echo pulse sample arrives Expand along the diagonal of the matrix, with each expanded element having a value of 0, meaning the dimension is 1. of Expanding to dimensions of Similarly, the dimension is of Expand the matrix diagonally to a dimension of of ; dimension is of Expanded to dimensions of ; dimension is of Expanded to dimensions of Its mathematical expression is as follows: Equation (42) In the formula, The dimension is The zero matrix, The dimension is The zero matrix, The dimension is The zero matrix, The dimension is The zero matrix, The dimension is The zero matrix, The dimension is The zero matrix; Based on the above explanation, equation (35) can be updated as follows: Equation (43) Therefore, with each online update traversing one echo sampling pulse, the target The dimension is increased by one order, and the update is performed within each unit echo region. After 1 data point The dimension will be from the initial The order increases to The order. It is worth noting that if the last unit echo region contains less than... Each data point is traversed directly using the same principle as described above.

[0077] Step 5: Full-scene image reconstruction based on global translation update For the recovery from each unit echo region A target grid, the edges of which The target grid is also affected by the echo regions of other units.

[0078] No. Each unit echo region is denoted as having a dimension of . of Therefore, it can be divided into ,in, , and It is of equal length The vector.

[0079] Similarly, from the first The target grid for recovering a unit echo region is: .Will Defined as overlapping displacement processing, it can be expressed as: Equation (44) Therefore, the global recursive update result for the complete echo region is: Equation (45) in, , This represents the floor function. This represents the summation operation for overlapping shift processing. Simultaneously, due to the repair of the antenna measurement matrix, a virtual target of length *l* exists at the edge; therefore, the final target reconstruction result... By truncation The length of the middle is It is obtained from a vector.

[0080] Step Six: Adaptive Selection of Regularization Parameters The above framework significantly reduces computational complexity and enables rapid imaging within an online framework. However, the imaging quality often depends heavily on the selection of the regularization parameter in equation (29). To avoid complex manual adjustment, an adaptive parameter selection method is proposed based on ridge estimation theory.

[0081] In classical statistics, In the context of regularization, the optimal generalized ridge parameter is calculated as follows: Equation (46) in, The variance can be estimated using the following formula: Equation (47) in, express The first in Each element value Let represent the mean. Based on the above formula, the regularization parameter corresponding to the optimal generalized ridge parameter can be obtained as: Equation (48) Thus, the regularization parameter of each local target mesh cell can be derived by mapping the target scattering coefficient value of the previous local target mesh cell, thereby achieving adaptive selection of the regularization parameter. The initial regularization value of the first edge local target mesh cell is iteratively initialized using the classic Tikhonov regularization method.

[0082] To verify the effectiveness of the method of the present invention, simulation results are as follows: Figures 5 to 12 As shown, the scanning area is set to... The original scene is as follows: Figure 5 As shown, it consists of three groups of adjacent targets at 3086m, 3124m, and 3161m, and two island-shaped targets at 3049m and 3199m. The azimuth interval between the adjacent targets at 3086m is... The target azimuth spacing at 3124m is The target azimuth spacing at 3161m is .

[0083] Real beam echo Figure 6 As shown, the real beam echo resolution is low, making it difficult to distinguish adjacent targets. After the target reconstruction algorithm, as... Figure 7 As shown, the TSVD algorithm improves resolution to some extent, but its high sidelobes severely affect image quality. Figure 8 The results of the Lucy algorithm shown indicate that the target cannot be reconstructed at all. Figure 9 The results of the IAA algorithm are presented, demonstrating good reconstruction performance. However, the resulting image shows some shadows. Figure 10 In the middle, from batch processing sparsity The regularization algorithm produces excellent resolution, but there is some blurring between closely adjacent targets. Figure 11 and Figure 12 Online sparse The processing results of the regularization algorithm and the algorithm proposed in this invention show that both algorithms achieve a high level of resolution enhancement, but... Figure 11 Some targets in the image exhibited defocusing, while Figure 12 The improved target reconstruction results demonstrate that the method of this invention can effectively reduce storage requirements while maintaining the advantages of online frames, thus achieving rapid real-time imaging.

[0084] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An online adaptive radar fast imaging method, characterized in that, Includes the following steps: Step S1: Establish an azimuth echo convolution model and repair the antenna measurement matrix to reduce imaging edge information errors; Step S2: Based on the independence of echo data within adjacent beamwidths and the structural characteristics of the repaired antenna measurement matrix, recursive updates of local beams are achieved. Step S3: Achieve global beam recursive update through overlap shifting processing; Step S4: Adaptively select regularization parameters based on the optimal generalized ridge estimation theory to complete the reconstruction of the target scene.

2. The online adaptive radar fast imaging method according to claim 1, characterized in that, In step S1, establishing the azimuth echo convolution model specifically includes: The first The echo data of each distance cell is modeled as follows: In the formula, Represents echo data, The target scattering coefficient, This represents zero-mean Gaussian white noise. Let represent the convolution matrix composed of antenna pattern samples, and . in, This is a sample of the antenna radiation pattern. The number of samples in the antenna pattern is denoted as , and ,in, For beamwidth, The pulse repetition frequency, This represents the antenna scanning speed.

3. The online adaptive radar fast imaging method according to claim 2, characterized in that, In step S1, the repair process for the antenna measurement matrix specifically includes: convolution matrix Repair to : Based on the repair Process it.

4. The online adaptive radar fast imaging method according to claim 3, characterized in that, In step S2, the recursive update of the local beam specifically includes: The full echo region is divided into several unit echo regions; When the When each echo pulse sample arrives, for Perform a windowing operation to extract non-zero elements, where... It is a convolution matrix After being divided into rows, the first OK; The corresponding parameter matrix is ​​expanded along its diagonal, with the expanded elements having a value of zero.

5. The online adaptive radar fast imaging method according to claim 4, characterized in that, The dimension change of the target vector update within a unit echo region satisfies: Each online update iterates through an echo sampling pulse, increasing the dimension of the target vector by one order, and updating within each unit echo region. After [number] data points, the dimension of the target vector increases to [number]. Rank.

6. The online adaptive radar fast imaging method according to claim 5, characterized in that, If the last unit echo region contains less than The data points were also traversed directly.

7. The online adaptive radar fast imaging method according to claim 1, characterized in that, In step S3, the global beam recursive update specifically includes: The target grid for restoring the i-th unit echo region. Divide into 3 equal-length subvectors: ,in, , and If they are subvectors of equal length, then the first... The target grid for recovering a unit echo region is: Then we have: ,in, This indicates overlapping displacement processing. express target vector and The target result superimposed after the target vector is shifted and translated by point L; The global recursive update result for the complete echo region is: In the formula, , This represents the number of azimuth sampling points for the target scene. The number of samples in the antenna pattern. It is the floor function; This represents the summation operation for overlapping displacement processing.

8. The online adaptive radar fast imaging method according to claim 7, characterized in that, Final goal: reconstruction result By truncation The length of the middle is It is obtained by using a vector.

9. The online adaptive radar fast imaging method according to claim 1, characterized in that, In step S4, the regularization parameter The adaptive selection calculation specifically includes: Calculate the regularization parameters of the current local target mesh cell based on the previous local target mesh. : In the formula, This indicates the number of samples in the antenna radiation pattern. Indicates the first The target scattering coefficient value of a local target grid cell. express The The average value of the target scattering coefficient of each local target grid cell.

10. The online adaptive radar fast imaging method according to claim 9, characterized in that, The regularization parameters of the first edge local target mesh element are obtained through iterative initialization using the Tikhonov regularization method.

Images