Adaptive Target Detection Method and Apparatus Based on Scattering Center Estimation

Abstract

This invention discloses an adaptive detection method and apparatus for range-extended targets based on scattering center estimation. A binary hypothesis testing model for range-extended targets is established, and the probability density functions under two hypotheses are determined. Maximum likelihood calculation is performed on the unknown parameters under both hypotheses to obtain the adaptive matched filter test statistic of the composite Gaussian model. The texture components of the detection unit are estimated using the correlation of texture components in the range dimension, yielding the test statistic after texture component estimation. A sparse regularized optimization model is constructed and solved to obtain a range-extended target detector with adaptive scattering center estimation. This method adaptively estimates the scattering center without requiring prior information about the target, reducing interference from clutter in units without a target scattering center. Furthermore, this method estimates the texture components based on the range-dimensional correlation of clutter, resulting in superior detection performance compared to traditional adaptive detection methods.

Description

Technical Field

[0001] This invention belongs to the field of adaptive radar signal detection, specifically relating to an adaptive detection method and apparatus for range-extended targets based on scattering center estimation. Background Technology

[0002] Radar, short for Radio Detection and Ranging, is an electronic device that uses the emission of electromagnetic waves and the reception of their reflected signals to detect and locate targets. Radar can detect long-range targets around the clock and in all weather conditions, playing a significant role in both military and civilian applications.

[0003] Radar adaptive detection is an important branch of radar signal processing. Its applications cover civilian and military fields such as air traffic control, autopilot, enemy target early warning, and target unknown speed acquisition, playing an irreplaceable role. However, with the increasing complexity of the electromagnetic environment, radar faces serious problems such as electronic interference and clutter interference. Therefore, the development of more advanced adaptive detection technology is particularly important.

[0004] Radar detects targets by receiving electromagnetic waves reflected from them. When the centers of the scattered points generated by a target are distributed within different range resolution cells of the radar, the target can be considered a range-extended target. Compared to point targets, the radar echoes of range-extended targets contain more target scattering centers. These scattering centers are related to the target's shape, size, and motion characteristics, containing richer target features. Many scholars both domestically and internationally have studied the adaptive detection problem of range-extended targets. Gerlach et al. proposed a Density Dependent General Likelihood Ratio Test (SDD-GLRT) detector and a Non-scatterer density Dependent General Likelihood Ratio Test (NSDD-GLRT) detector for Spherically Invariant Random Vector (SIRV) clutter environments. Simulation experiments demonstrate that in scenarios where the distribution of target scattering centers is relatively sparse, the SDD-GLRT detector exhibits superior performance compared to the NSDD-GLRT detector. To address the issue that target scattering centers typically occupy only a portion of the range cells in the detection data, You He et al. proposed an Order Statistics GLRT (OS-GLRT) detector and a Dynamic Threshold OS-GLRT (DOS-GLRT) detector by selecting range cells with larger test statistics for detection, thus improving the robustness of the detectors. However, these methods rely on prior information about the number of scattering points in the detection data. In practical problems, the number of target scattering centers is often unknown, which limits the performance of these methods in real-world scenarios.

[0005] Therefore, there is an urgent need for an adaptive detection method that can improve the detection performance of range-extended targets against a background of complex Gaussian clutter. Summary of the Invention

[0006] Purpose of the invention: This invention proposes a range-extended adaptive target detection method and device based on target scattering center estimation. It constructs a hypothesis testing model under composite Gaussian clutter, estimates texture components through the range dimension correlation of clutter, and estimates the number and location of scattering centers in the detection area by combining sparse regularization, thereby improving the detection performance of the detector.

[0007] Technical solution: The present invention provides a range-extended adaptive target detection method based on target scattering center estimation, comprising the following steps:

[0008] (1) Establish a binary hypothesis testing model for the distance extension target and determine the probability density function under the two hypotheses;

[0009] (2) The maximum likelihood of the unknown parameters in the two hypotheses is solved to obtain the adaptive matched filter test statistic of the composite Gaussian model;

[0010] (3) The correlation of texture components in the distance dimension is used to estimate the texture components of the unit to be detected, and the test statistic after the texture component estimation is obtained.

[0011] (4) Take the logarithm of the test statistic in step (3) and construct a sparse regularized optimization model;

[0012] (5) Solve the sparse regularized optimization model to obtain the range-extended target detector with adaptive estimation of scattering center.

[0013] Furthermore, the process of establishing the distance extension target binary hypothesis testing model in step (1) is as follows:

[0014] Assuming radar transmission The received echo data includes (pulse, pulse ... ) distance units are used to establish a binary hypothesis testing model for the distance expansion target:

[0015]

[0016] in, and These represent the null hypothesis and the alternative hypothesis, respectively. Indicates the first Echo data from each unit, Indicates the first Clutter data of each unit, Indicates the first The strength of the target signal in each unit represents the target guidance. directional vector, The number of auxiliary data reference units, The number of main data detection units.

[0017] Furthermore, the process of determining the probability density function under the two assumptions in step (1) is as follows:

[0018] In the detection area, the data to be detected under two hypotheses The conditional probability density function is expressed as:

[0019]

[0020]

[0021] in, and They represent the first and second hypotheses, respectively. Texture components of clutter per range unit A vector representing texture components at different distance units. A vector representing the amplitude of targets at different distance units. The time-dimensional covariance matrix of clutter is represented. and These represent the transpose and conjugate transpose operations, respectively.

[0022] The texture components of clutter are modeled as a multidimensional inverse Gaussian distribution, with the following probability density function:

[0023]

[0024] in, and These are the shape and scale parameters of the inverse Gaussian distribution, respectively. , Let be the range covariance matrix of the clutter.

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

[0026] Based on the generalized likelihood ratio test criterion, the GLRT test statistic for the generalized likelihood ratio test under two hypotheses is obtained:

[0027]

[0028] A two-step method is used to simplify the GLRT test statistic. First, the texture components are assumed. and Given the known values, and then replacing them with their estimated values, we obtain a suboptimal adaptive detector:

[0029]

[0030] Taking the logarithm of the conditional probability density function of the data to be tested under the alternative hypothesis, we have:

[0031] Regarding the formula Find the partial derivative, where This indicates the conjugate operation, and setting the partial derivatives to zero, we obtain... Assuming Maximum likelihood estimation:

[0032] ;

[0033] The constrained approximate maximum likelihood AML estimation method is used to estimate the covariance matrix. Make an estimate:

[0034]

[0035] in, Represents the matrix trace operation. Indicates the first The estimated value of the covariance matrix obtained in the second iteration initial value Represented as:

[0036] .

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

[0038] Using auxiliary data texture components to estimate the texture components of the unit to be detected, a length of ( ) window, make , , The posterior probability density function is:

[0039]

[0040] Using the maximum a posteriori probability method Seeking information about The maximum values ​​are:

[0041]

[0042]

[0043] Move the window and repeat the above steps to obtain the texture component estimate for each detection unit;

[0044] By substituting the estimates of target intensity, covariance matrix, and texture components into the adaptive detector, the following results are obtained:

[0045] .

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

[0047] Taking the logarithm of the test statistic obtained in step (3), we have:

[0048]

[0049] Detection area Decomposed into target components and clutter components The main part of the test statistic is expressed as follows:

[0050] Equivalent representation in vector form:

[0051]

[0052]

[0053] in, Represents the identity matrix. , , The sparse regularization optimization model is constructed as follows:

[0054]

[0055] in, and Let represent the 0-norm and 2-norm of the vectors, respectively; using the Lagrange multiplier method, it is transformed into an unconstrained optimization problem:

[0056]

[0057] in, Represents the loss function. This represents the regularization parameter.

[0058] Furthermore, the implementation process of step (5) is as follows:

[0059] (51) Initialize parameters: Initialize the original data to be estimated. Elements in descending order , Regularization parameters Regularization parameter correction factor , Maximum number of iterations Initial iteration count Output results ;

[0060] (52) Start iterating when hour:

[0061]

[0062]

[0063]

[0064]

[0065] (53) When season Repeat step (52);

[0066] (54) Iteration complete. The element in the middle is the estimated scattering center. China retains The elements in the array and their positions are set, and the remaining elements are set to zero. This is then assigned to... ,remember for The set of indices of non-zero elements;

[0067] (55) The range-extended target detector based on scattering center estimation is:

[0068] .

[0069] The present invention provides a storage medium storing a computer program, which, when executed by at least one processor, implements the steps of the range-extended target adaptive detection method based on scattering center estimation as described above.

[0070] An electronic device according to the present invention includes a memory and a processor, wherein:

[0071] Memory is used to store computer programs that can run on a processor;

[0072] The processor is configured to, while running the computer program, execute the steps of the range-extended target adaptive detection method based on scattering center estimation as described above.

[0073] Beneficial Effects: Compared with existing technologies, the present invention offers the following advantages: For the adaptive detection of range-extended targets against a composite Gaussian distribution background with inverse Gaussian texture, the present invention proposes a novel detection method. It uses the maximum a posteriori probability criterion to estimate texture components based on the range-dimensional correlation of clutter. Finally, a sparse optimization model is constructed, and sparse regularization is used to adaptively estimate the number and location of range-extended target scattering points in the detection region. Compared with traditional adaptive detection methods for range-extended targets, the present invention does not require prior information about the target scattering center and can adaptively estimate the scattering center, reducing interference from clutter units without a target scattering center. Furthermore, the method estimates texture components based on the range-dimensional correlation of clutter, resulting in superior detection performance compared to traditional adaptive detection methods. Attached Figure Description

[0074] Figure 1 This is a flowchart of the method of the present invention;

[0075] Figure 2 The figure shows the effect of different covariance matrix parameters on the false alarm probability of the proposed detector.

[0076] Figure 3 Figure 1 shows the effect of different regularization parameters on the detection performance of the proposed detector.

[0077] Figure 4 A comparison chart of the detection performance of the proposed detector with other detectors; Detailed Implementation

[0078] The present invention will now be described in further detail with reference to the accompanying drawings.

[0079] like Figure 1 As shown, this invention proposes a range-extended adaptive target detection method based on target scattering center estimation, and the specific implementation steps are as follows:

[0080] Step 1: Establish a binary hypothesis model and determine the probability density function under the two hypotheses.

[0081] Step 1-1, assuming radar transmission Each pulse, the received echo data contains Using distance units, establish a binary hypothesis testing model for the distance expansion target:

[0082]

[0083] in and These represent the null hypothesis and the alternative hypothesis, respectively. Indicates the first Echo data from each unit, Indicates the first Clutter data of each unit, Indicates the first The intensity of the target signal in each unit, Indicates the target guidance vector. The number of auxiliary data reference units, The number of main data detection units.

[0084] Steps 1-2: In the detection area, the data to be detected under two hypotheses. The conditional probability density function is expressed as:

[0085]

[0086]

[0087] in, and They represent the first under two different assumptions. Texture components of clutter per range unit A vector representing texture components at different distance units. A vector representing the amplitude of targets at different distance units. The time-dimensional covariance matrix of clutter is represented. and These represent the transpose and conjugate transpose operations, respectively.

[0088] The texture components of clutter are modeled as a multidimensional inverse Gaussian distribution, with the following probability density function:

[0089]

[0090] in and These are the shape and scale parameters of the inverse Gaussian distribution, respectively. , Let be the range covariance matrix of the clutter.

[0091] Step 2: Perform maximum likelihood calculation on the unknown parameters in both hypotheses to obtain the Adaptive Matched Filter (AMF) test statistic for the composite Gaussian model.

[0092] Step 2-1: Based on the generalized likelihood ratio test criterion, the generalized likelihood ratio (GLRT) test statistics under the two hypotheses can be obtained:

[0093]

[0094] A two-step method is used to simplify the GLRT test statistic. First, the texture components are assumed. and Given the known values, and then replacing them with their estimated values, we can obtain a suboptimal adaptive detector:

[0095] .

[0096] Step 2-2, take the contents of step 1-2 Taking the logarithm, we have:

[0097]

[0098] Regarding the formula Find the partial derivative, where The conjugate operation is represented by the expression, and by setting the partial derivatives to zero, we can obtain the following: Assuming Maximum likelihood estimation:

[0099] .

[0100] Steps 2-3 involve using the constrained approximate maximum likelihood (AML) estimation method to estimate the covariance matrix. Make an estimate:

[0101]

[0102] in Represents the matrix trace operation. Indicates the first The estimated value of the covariance matrix obtained in the second iteration initial value Represented as:

[0103] .

[0104] Step 3: Estimate the texture components of the unit to be detected by utilizing the correlation of texture components in the distance dimension.

[0105] Step 3-1: Use auxiliary data texture components to estimate the texture components of the unit to be detected, taking a length of... The window, , , The posterior probability density function is:

[0106]

[0107] Using the maximum a posteriori probability method Seeking information about The maximum values ​​are:

[0108]

[0109]

[0110] Step 3-2: Move the window and repeat step 3-1 to obtain the texture component estimate for each detection unit.

[0111] Step 3-3: Substitute the estimated values ​​of target intensity, covariance matrix, and texture components into the adaptive detector from step 2-1 to obtain the results:

[0112] .

[0113] Step 4: Take the logarithm of the test statistic in Step 3 to construct a sparse regularized optimization model.

[0114] Step 4-1, taking the logarithm of the test statistic from step 3-3, we have:

[0115] .

[0116] Step 4-2, in the detection area Decomposed into target components and clutter components The main part of the test statistic can be expressed as:

[0117]

[0118] Equivalent representation in vector form:

[0119]

[0120]

[0121] in Represents the identity matrix. , , The optimization model is constructed as follows:

[0122]

[0123] in and Let represent the 0-norm and 2-norm of the vector, respectively. Using the Lagrange multiplier method, we transform it into an unconstrained optimization problem:

[0124]

[0125] in Represents the loss function. This represents the regularization parameter.

[0126] Step 5: Solve the sparse regularized optimization model to obtain the range-extended target detector with adaptive estimation of the scattering center.

[0127] Step 5-1, Initialize parameters:

[0128] Original data Elements in descending order , Regularization parameters Regularization parameter correction factor , Maximum number of iterations Initial iteration count Output results .

[0129] Step 5-2, begin iteration, when hour,

[0130]

[0131]

[0132]

[0133]

[0134] Step 5-3, when season Repeat step 5-2.

[0135] Step 5-4, iteration complete. The element in the middle is the estimated scattering center. China retains The elements in the array and their positions are set, and the remaining elements are set to zero. This is then assigned to... ,remember for The set of indices of non-zero elements;

[0136] Step 5-5, the range-extended target detector based on the scattering center estimation is as follows:

[0137] .

[0138] 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 range-extended target adaptive detection method based on scattering center estimation as described above.

[0139] The present invention also provides an electronic device, including a memory and a processor, wherein: the memory is used to store a computer program that can run on the processor; the processor is used to execute, when running the computer program, the steps of the range-extended target adaptive detection method based on scattering center estimation as described above.

[0140] The invention is verified through simulation experiments and measured data experiments. In the experiments, the detection unit was set to 10, the number of target scattering centers was 4, and they were randomly distributed in the detection area. Monte Carlo simulation was used to comprehensively evaluate the performance of the proposed detector, with a false alarm rate set at [value missing]. Monte Carlo times First, the CFAR characteristics of the detector relative to clutter parameters are verified. Second, the impact of different regularization parameters on the detector performance is analyzed. Finally, the proposed detector is compared with the classic range-extended target detector to evaluate the detection performance of the proposed method.

[0141] Figure 2 To illustrate the impact of different clutter covariance matrix parameters on the false alarm probability of the proposed method, from... Figure 2 As can be seen, the false alarm rate remains basically unchanged as the parameters of the clutter covariance matrix change. Figure 3 To illustrate the impact of different regularization parameters on the detection probability of the proposed method, from... Figure 3As can be seen, the detection probability curve of the detector does not change with the regularization parameter. Therefore, the proposed method can adaptively detect targets with different scattering center distributions.

[0142] To verify the feasibility and effectiveness of the method of this invention, it is compared with three classic distance-extended target detection algorithms: SDD-GLRT, NSDD-GLRT, and OS-GLRT. The comparison results are shown in the figure below. Figure 4 As shown, both SDD-GLRT and OS-GLRT methods require prior information about the target scattering center, while NSDD-GLRT requires incoherent accumulation of the energy of all range cells in the detection region, thus being affected by clutter that does not contain range cells with scattering centers. The method proposed in this invention can adaptively estimate the location of the scattering center in the detection region and estimate the texture components based on the correlation of the range dimension of clutter, resulting in detection performance superior to traditional detection methods.

[0143] The above description is only a preferred embodiment of the present invention. It should be noted that those skilled in the art can make several improvements and modifications 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 range-extended target adaptive detection method based on scattering center estimation, characterized in that, Includes the following steps: (1) Establish a binary hypothesis testing model for the distance extension target and determine the probability density function under the two hypotheses; (2) The maximum likelihood of the unknown parameters in the two hypotheses is solved to obtain the adaptive matched filter test statistic of the composite Gaussian model; (3) The correlation of texture components in the distance dimension is used to estimate the texture components of the unit to be detected, and the test statistic after the texture component estimation is obtained. (4) Take the logarithm of the test statistic in step (3) and construct a sparse regularized optimization model; (5) Solve the sparse regularized optimization model to obtain the range-extended target detector with adaptive estimation of scattering center.

2. The adaptive target detection method based on scattering center estimation according to claim 1, characterized in that, The process of establishing the distance extension target binary hypothesis testing model in step (1) is as follows: Assuming radar launch The received echo data includes (pulse, pulse ... ) distance units are used to establish a binary hypothesis testing model for the distance expansion target: in, and These represent the null hypothesis and the alternative hypothesis, respectively. Indicates the first Echo data from each unit, Indicates the first Clutter data of each unit, Indicates the first The strength of the target signal in each unit represents the target guidance. directional vector, The number of secondary data reference units, The number of main data detection units.

3. The adaptive detection method for range-extended targets based on scattering center estimation according to claim 1, characterized in that, The process of determining the probability density function under the two hypotheses in step (1) is as follows: In the detection area, the data to be detected under two hypotheses The conditional probability density function is expressed as: in, and They represent the first and second hypotheses, respectively. Texture components of clutter per range unit A vector representing texture components at different distance units. A vector representing the amplitude of targets at different distance units. The time-dimensional covariance matrix of clutter is represented. and These represent the transpose and conjugate transpose operations, respectively. The texture components of clutter are modeled as a multidimensional inverse Gaussian distribution, with the following probability density function: in, and These are the shape and scale parameters of the inverse Gaussian distribution, respectively. , Let be the range covariance matrix of the clutter.

4. The adaptive detection method for range-extended targets based on scattering center estimation according to claim 1, characterized in that, The implementation process of step (2) is as follows: Based on the generalized likelihood ratio test criterion, the GLRT test statistic for the generalized likelihood ratio test under two hypotheses is obtained: A two-step method is used to simplify the GLRT test statistic. First, the texture components are assumed. and Given the known values, and then replacing them with their estimated values, we obtain a suboptimal adaptive detector: Taking the logarithm of the conditional probability density function of the data to be tested under the alternative hypothesis, we have: Regarding the formula Find the partial derivative, where This indicates the conjugate operation, and setting the partial derivatives to zero, we obtain... Assuming Maximum likelihood estimation: ； The constrained approximate maximum likelihood AML estimation method is used to estimate the covariance matrix. Make an estimate: in, Represents the matrix trace operation. Indicates the first The estimated value of the covariance matrix obtained in the second iteration initial value Represented as: 。 5. The adaptive detection method for range-extended targets based on scattering center estimation according to claim 1, characterized in that, The implementation process of step (3) is as follows: Auxiliary data texture components are used to estimate the texture components of the unit to be detected, taking a length of ( ) window, make , , The posterior probability density function is: Using the maximum a posteriori probability method Seeking information about The maximum values ​​are: Move the window and repeat the above steps to obtain the texture component estimate for each detection unit; By substituting the estimates of target intensity, covariance matrix, and texture components into the adaptive detector, the following results are obtained: 。 6. The adaptive target detection method based on scattering center estimation according to claim 1, characterized in that, The implementation process of step (4) is as follows: Taking the logarithm of the test statistic obtained in step (3), we have: Detection area Decomposed into target components and clutter components The main part of the test statistic is expressed as follows: Equivalent representation in vector form: in, Represents the identity matrix. , , The sparse regularization optimization model is constructed as follows: in, and Let represent the 0-norm and 2-norm of the vectors, respectively; using the Lagrange multiplier method, it is transformed into an unconstrained optimization problem: in, Represents the loss function. This represents the regularization parameter.

7. The adaptive detection method for range-extended targets based on scattering center estimation according to claim 1, characterized in that, The implementation process of step (5) is as follows: (51) Initialize parameters: Initialize the original data to be estimated. Elements in descending order , Regularization parameters Regularization parameter correction factor , Maximum number of iterations Initial iteration count Output results ; (52) Start iterating when hour: (53) When season Repeat step (52); (54) Iteration complete. The element in the middle is the estimated scattering center. China retains The elements in the array and their positions are set, and the remaining elements are set to zero. This is then assigned to... ,remember for The set of indices of non-zero elements; (55) The range-extended target detector based on scattering center estimation is: 。 8. 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 range-extended target adaptive detection method based on scattering center estimation as described in any one of claims 1 to 7.

9. An electronic 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 range-extended target adaptive detection method based on scattering center estimation as described in any one of claims 1 to 7.

Images