Ground target detection method, system and electronic equipment for enhancing clutter edge performance

Abstract

This application belongs to the field of ground radar digital signal processing technology, and provides a ground target detection method, system, and electronic device to enhance clutter edge performance. The detection method includes: pre-setting that at most one clutter edge exists within a reference window of the radar echo data samples; determining the clutter edge position using clutter edge position estimates and selecting a retained sample set; calculating weighting coefficients by combining distribution parameter estimates and standardized residuals; constructing an objective function using the negative likelihood estimation function of the samples; solving for outlier vectors; constructing an index function based on the outlier vectors to filter outliers and obtain the zero norm of the index function; calculating the background clutter power estimate of the samples based on the outlier vector index function and the zero norm of the index function to obtain the corresponding detection threshold; and determining the target based on the detection threshold. This application can effectively avoid the target masking effect caused by strong clutter, and improves false alarm control capability while enhancing clutter edge detection performance.

Description

Technical Field

[0001] This application relates to the field of digital signal processing technology for ground radar, specifically to a ground target detection method, system, and electronic device that enhances clutter edge performance. Background Technology

[0002] Automatic target detection (ADR) of radar calculates a dynamic threshold value by analyzing echo data in real time and compares it with the actual echo signal to determine the presence of a target. This method can dynamically adjust the detection threshold according to environmental changes, effectively maintaining a constant false alarm rate (CFAR) even under conditions of constantly changing clutter power. Existing CFAR detection methods are mostly designed for specific interference environments and are difficult to adapt to complex and changing application environments. Mean-based CFAR performs poorly in non-uniform environments, ordered statistical CFAR struggles to maintain a low false alarm probability under strong clutter interference, and adaptive CFAR has high computational complexity, making it difficult to meet the real-time requirements of radar systems.

[0003] As radar applications expand, their operating environments and interference clutter become increasingly complex and diverse. Firstly, complex terrain or topographical variations lead to the coexistence of strong and weak clutter regions, causing significant variations in average clutter power among adjacent range cells. This non-uniformity significantly reduces target detection stability; when a weak target is adjacent to strong clutter, the target detection probability drops drastically, resulting in target masking. Secondly, fixed detection threshold methods struggle to adapt to fluctuations in the signal-to-clutter ratio within the transition zone between strong and weak clutter, leading to an increased false alarm probability. Finally, in non-uniform clutter environments, to ensure the real-time operation of the radar system, the detection threshold needs timely adjustment to adapt to clutter power changes, necessitating improved computational efficiency of the detection algorithm. Therefore, improving detection performance in complex environments while maintaining a constant false alarm rate has become a key challenge for radar target detection. Thus, it is necessary to propose a new method that not only improves the robustness of target detection, maintains constant false alarm characteristics, and enhances the real-time performance of the algorithm in clutter-edge environments, but also holds significant importance for researching target detection technology in non-uniform clutter environments for ground-based radar. Summary of the Invention

[0004] To address one of the aforementioned technical deficiencies, this application provides a ground target detection method, system, and electronic device that enhances clutter edge performance.

[0005] According to the first aspect of this application, a ground target detection method with enhanced clutter edge performance is provided, comprising: There is at most one clutter edge within the reference window of the preset radar echo data sample. The clutter edge position is determined by the clutter edge position estimate, and the retained sample set is selected based on the clutter edge position. By combining the estimated distribution parameters with the weighting coefficients obtained from the standardized residuals, the objective function is constructed using the negative likelihood estimation function of the sample. The outlier vector is then obtained by solving the outlier vector index function constructed from the outlier vector. Outlier values ​​are then screened based on the outlier vector index function to obtain the zero norm of the index function. The background clutter power estimate of the current sample is calculated by using the outlier vector index function and the zero norm of the index function, and the corresponding detection threshold is obtained. Target determination is then performed based on the detection threshold.

[0006] Preferably, the step of determining the clutter edge position through the clutter edge position estimate and selecting the retained sample set based on the clutter edge position specifically includes: Assuming that clutter edges exist in the samples within the reference window, the log-likelihood function is constructed based on the probability density function of the Weibull distribution using the sample data on both sides of the clutter edge, and the estimated clutter parameters are calculated. Calculate the margin test statistic based on the obtained clutter parameter estimates; The existence of clutter edges is determined by the obtained edge test statistics. If no clutter edges exist, the hypothesis is not valid, and the entire sample set is selected and retained. If clutter edges exist, the estimated location of the clutter edges is calculated. The clutter edge location is determined based on the obtained clutter edge location estimate, and a retained one-sided sample set is selected.

[0007] Preferably, the step of combining the weighted coefficients obtained from the distribution parameter estimates and standardized residuals, constructing an objective function using the negative likelihood estimation function of the samples, solving for the outlier vector, and then filtering outliers based on the outlier vector index function to obtain the zero norm of the index function specifically includes: Transform the samples within the reference window into a Gumbel distribution, and use the Gumbel distribution parameters to calculate the sample mean and standard deviation to obtain the estimated Gumbel distribution parameters. The weights are calculated using the estimated distribution parameters and the standardized residuals obtained by definition, resulting in the weighting coefficients. Construct a negative likelihood estimation function using the retained one-sided samples and the parameters of the Gumbel distribution; The objective function is constructed based on the weighting coefficients and the negative likelihood estimation function, and the outlier vector is obtained by solving it. An outlier index function is constructed using outlier vectors. Based on the outlier index function, outlier values ​​are filtered according to the values ​​of the outlier vectors to obtain the zero norm of the index function.

[0008] Preferably, the probability density function based on the Weibull distribution constructs a log-likelihood function using sample data from both sides of the clutter edge to calculate clutter parameter estimates; the edge test statistic is calculated based on the obtained clutter parameter estimates, specifically including: The total length of the sample within the preset reference window is 2. N Assuming the clutter edge appears in the first... M The first reference unit and the first M Between +1 reference units, where M The range of values ​​is , to obtain the preceding M The sample set y of each sample and the sample set after inclusion The sample set z of samples; Based on the probability density function of the Weibull distribution, a log-likelihood function is constructed using the two sample sets y and z. Log-likelihood estimation is then performed on the unknown parameters. By solving the nonlinear equation of the log-likelihood estimation, the estimated scale parameter of the sample is obtained. and Shape parameter estimates and ; Based on the obtained scale parameter estimates and Calculate the marginal test statistic T C :

[0009] in, and These are the estimated variances of the sample parameters on both sides.

[0010] Preferably, the step of determining whether a clutter edge exists based on the obtained edge test statistic, and if no clutter edge exists, the hypothesis is not valid, and the entire sample set is selected and retained; if a clutter edge exists, the estimated clutter edge position is calculated; and the clutter edge position is determined based on the obtained estimated clutter edge position and the selected retained one-sided sample set, specifically includes: The obtained margin test statistic T C The critical value of the pre-set test statistic T crit Compared to determining whether clutter edges exist, when When the clutter distribution is uniform and has no clutter edges, the assumption is invalid, and all samples x are selected and retained; when At that time, the clutter on both sides differed significantly, and the clutter distribution had clutter edges. The estimated location of the clutter edges was calculated. ; Based on clutter edge location estimation Determine the clutter edge location and select the retained one-sided sample set u, when The clutter edge appears to the right of sample x, taking u=y; when The clutter edge appears to the left of sample x, so we take u=z; Among them, the clutter edge location estimate The total log-likelihood value is calculated as follows:

[0011] in, This represents the total log-likelihood value.

[0012] Preferably, the step of calculating the sample mean and standard deviation using the Gumbel distribution parameters to obtain Gumbel distribution parameter estimates; calculating weights using the distribution parameter estimates and the standardized residuals defined by the definition to obtain weighting coefficients; constructing a negative likelihood estimation function using the retained one-sided samples and the Gumbel distribution parameters; constructing an objective function based on the weighting coefficients and the negative likelihood estimation function; and solving for the outlier vector specifically includes: The scale parameter of the Gumbel distribution is obtained from the Gumbel distribution. and position parameters The estimated parameters of the Gumbel distribution are obtained by calculating the sample mean and standard deviation. and By defining each sample relative to the position parameter Standardized residuals Specifically:

[0013] Based on standardized residuals Calculate the weights to obtain the weighting coefficients. Specifically:

[0014] The negative likelihood estimation function for one-sided samples is obtained through calculation. Specifically:

[0015] in, K It is the number of samples in the one-sided sample set u; The objective function for outlier detection is constructed based on the negative likelihood estimation function, specifically as follows:

[0016] in, For outlier vectors, , For outlier parameters, The outlier is defined as follows: .

[0017] Preferably, the step of constructing an outlier index function using outlier vectors, and then filtering outliers based on the values ​​of the outlier vectors to obtain the zero norm of the index function, specifically includes: Solving for outlier vectors by minimizing the objective function yields the optimization problem for non-convex outlier vectors:

[0018] Using Taylor's formula for single variables Construct an upper bound function for each single variable. The optimal solution is:

[0019] Among them, operators This operator ensures ;when When, the parameters are obtained. The closed-form solution is as follows:

[0020] in, This indicates the probability of classifying a normal data sample as an outlier. Given all relevant parameters, solve for the outlier vector o. Based on the value of outlier vector o, perform outlier filtering to obtain the outlier index function and the zero norm of the index function, specifically:

[0021] in, The outlier index function is represented by the index function. This represents the zero norm of the index function.

[0022] Preferably, the step of calculating the background clutter power estimate of the current sample using the outlier vector index function and the zero norm of the index function to obtain the corresponding detection threshold, and then determining the target based on the detection threshold, specifically includes: Background clutter power estimates are calculated based on the outlier vector index function and the zero norm of the index function. Specifically:

[0023] Estimated background clutter power Multiplying the result by a preset threshold coefficient yields the corresponding detection threshold, specifically:

[0024] in, T Represents the threshold factor; By comparing the target amplitude D of the unit to be detected with the corresponding detection threshold S, if If the point is determined to be the target point, then the target point is determined to be the target point; otherwise, the target does not exist.

[0025] According to a second aspect of this application, a ground target detection system with enhanced clutter edge performance is provided, including a module for implementing the ground target detection method with enhanced clutter edge performance.

[0026] According to a third aspect of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the ground target detection method with enhanced clutter edge performance.

[0027] The target detection method provided in this application introduces clutter edge determination and outlier screening to design an efficient and accurate ground radar target detection method. It aims to improve radar target detection performance in complex environments while maintaining a constant false alarm rate. First, it presupposes the existence of clutter edges. At each possible clutter edge location, it sequentially calculates the estimated value of background clutter parameters. Based on the parameter estimates, it constructs an edge test statistic. The existence of a clutter edge is determined based on the edge test statistic. If no clutter edge is found, all samples are retained. If a clutter edge is found, the estimated clutter edge location is calculated. Based on the estimated clutter edge location, the retained one-sided samples are selected for the next calculation. The clutter edge determination introduced in this application improves the detection performance in clutter-edge environments and enhances the computational efficiency and applicability of the algorithm. Then, it calculates weighting coefficients based on the distribution parameter estimates. An objective function is constructed using the weighting coefficients and the negative likelihood estimation function. The outlier vector is obtained by solving for the objective function, and an index function is constructed using the outlier vector for outlier screening. The outlier screening introduced in this application not only improves the reliability of subsequent clutter power estimation but also reduces the constant false alarm rate loss of the algorithm. Finally, the background clutter power estimate of the current sample is calculated based on the outlier vector index function and the zero norm of the index function to obtain the detection threshold. The amplitude of the target cell is compared with the detection threshold to determine the target. The target detection method provided in this application effectively improves the robustness of target detection, maintains constant false alarm rate characteristics, and enhances the efficiency and real-time performance of the algorithm in non-uniform clutter environments. It can effectively avoid the target occlusion effect caused by strong clutter, and achieves improved false alarm control capability while enhancing clutter edge detection performance.

[0028] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of what is pointed out in the written description and the accompanying drawings. Attached Figure Description

[0029] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 The signal processing flowchart provided for this application.

[0030] Figure 2 A flowchart of clutter edge localization processing provided in this application.

[0031] Figure 3 The flowchart for weighted outlier detection processing provided in this application.

[0032] Figure 4 The flowchart of the target detection method provided in this application.

[0033] Figure 5 The detection threshold diagram of this invention is provided for when the target is close to the edge of clutter.

[0034] Figure 6 This invention provides a detection threshold diagram for when the target is far from the edge of clutter.

[0035] Figure 7 A comparison chart of detection probabilities under clutter edges provided in this application.

[0036] Figure 8 A comparison chart of false alarm probabilities under clutter edge conditions provided in this application. Detailed Implementation

[0037] To make the technical solutions and advantages of the embodiments of this application clearer, the exemplary embodiments of this application will be described in further detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0038] Addressing some problems existing in the current technology: In a first aspect, embodiments of this application provide a ground target detection method with enhanced clutter edge performance. This method can be executed by a ground target detection device with enhanced clutter edge performance, or by components configured within the target detection device, such as processors, chips, chip systems, etc., or by logic modules or software having some or all of the functions of the target detection device. This application does not limit this aspect.

[0039] For example, this ground target detection method with enhanced clutter edge performance, such as Figure 1-4 As shown, the target detection method includes: There is at most one clutter edge within the reference window of the preset radar echo data sample. The clutter edge position is determined by the clutter edge position estimate, and the retained sample set is selected based on the clutter edge position. By combining the estimated distribution parameters with the weighting coefficients obtained from the standardized residuals, the objective function is constructed using the negative likelihood estimation function of the sample. The outlier vector is then obtained by solving the outlier vector index function constructed from the outlier vector. Outlier values ​​are then screened based on the outlier vector index function to obtain the zero norm of the index function. The background clutter power estimate of the current sample is calculated by using the outlier vector index function and the zero norm of the index function, and the corresponding detection threshold is obtained. Target determination is then performed based on the detection threshold.

[0040] Specifically, this application discloses a ground target detection method that enhances clutter edge performance for target detection in non-uniform clutter environments of ground radar. The radar received echo is down-converted, decimation filtered, pulse compressed, and coherently accumulated to form range-Doppler two-dimensional data. In order to obtain the range and Doppler information of the target, this application uses the range dimension data in the range-Doppler two-dimensional data as the input sample set, and introduces clutter edge judgment and outlier screening to design this efficient and accurate ground radar target detection method.

[0041] In some possible ways of implementing the first aspect, such as Figure 1-2 As shown, the step of determining the clutter edge position through the clutter edge position estimate and selecting the retained sample set based on the clutter edge position specifically includes: Assuming that clutter edges exist in the samples within the reference window, the log-likelihood function is constructed based on the probability density function of the Weibull distribution using the sample data on both sides of the clutter edge, and the estimated clutter parameters are calculated. Calculate the margin test statistic based on the obtained clutter parameter estimates; The existence of clutter edges is determined by the obtained edge test statistics. If no clutter edges exist, the hypothesis is not valid, and the entire sample set is selected and retained. If clutter edges exist, the estimated location of the clutter edges is calculated. The clutter edge location is determined based on the obtained clutter edge location estimate, and a retained one-sided sample set is selected.

[0042] In some possible ways of implementing the first aspect, such as Figure 3 As shown, the weighted coefficients obtained by combining the estimated distribution parameters and standardized residuals are used to construct the objective function using the negative likelihood estimation function of the samples. The outlier vector is then solved, and the outlier values ​​are filtered based on the outlier vector index function constructed from the outlier vectors to obtain the zero norm of the index function. Specifically, this includes: Transform the samples within the reference window into a Gumbel distribution, and use the Gumbel distribution parameters to calculate the sample mean and standard deviation to obtain the estimated Gumbel distribution parameters. The weights are calculated using the estimated distribution parameters and the standardized residuals obtained by definition, resulting in the weighting coefficients. Construct a negative likelihood estimation function using the retained one-sided samples and the parameters of the Gumbel distribution; The objective function is constructed based on the weighting coefficients and the negative likelihood estimation function, and the outlier vector is obtained by solving it. An outlier index function is constructed using outlier vectors. Based on the outlier index function, outlier values ​​are filtered according to the values ​​of the outlier vectors to obtain the zero norm of the index function.

[0043] In some possible implementations of the first aspect, the probability density function based on the Weibull distribution constructs a log-likelihood function using sample data from both sides of the clutter edge to calculate clutter parameter estimates; the edge test statistic is calculated based on the obtained clutter parameter estimates, specifically including: The total length of the sample within the preset reference window is 2. N Assuming the clutter edge appears in the first... M The first reference unit and the first M Between +1 reference units, where M The range of values ​​is , to obtain the preceding M The sample set y of each sample and the sample set after inclusion The sample set z of samples; Based on the probability density function of the Weibull distribution, a log-likelihood function is constructed using the two sample sets y and z. Log-likelihood estimation is then performed on the unknown parameters. By solving the nonlinear equation of the log-likelihood estimation, the estimated scale parameter of the sample is obtained. and Shape parameter estimates and ; Based on the obtained scale parameter estimates and Calculate the marginal test statistic T C :

[0044] in, and These are the estimated variances of the sample parameters on both sides.

[0045] Specifically, the radar echo is down-converted, decimation filtered, pulse compressed, and coherently accumulated to form range-Doppler two-dimensional data. For the range dimension, the total length is 2. N The reference window sample x, assuming the clutter edge appears in the first... M The first reference unit and the first M Between +1 reference units, where M The range of values ​​is , to obtain the preceding M The sample set y of each sample and the sample set after inclusion The sample sets z of two samples are independent and identically distributed. They can be used to construct a log-likelihood function to estimate the unknown parameters. By using Newton's method to define the objective function and convergence conditions, the nonlinear equation of the log-likelihood estimation is solved to obtain the estimated values ​​of the scale parameter and shape parameter of the two samples.

[0046] Furthermore, the embodiments of this application make full use of every piece of collected running data during the data processing division and sample set extraction process. Compared with traditional data averaging operations, this ensures the speed and effectiveness of data calculation and improves the accuracy and comprehensiveness of the final detection results.

[0047] In some possible implementations of the first aspect, the determination of whether a clutter edge exists based on the obtained edge test statistic, and if no clutter edge exists, the hypothesis is not valid, and the entire sample set is selected and retained; if a clutter edge exists, an estimated clutter edge location is calculated; and the clutter edge location is determined based on the obtained estimated clutter edge location and the retained one-sided sample set is selected, specifically including: The obtained margin test statistic T C The critical value of the pre-set test statistic T crit Compared to determining whether clutter edges exist, when When the clutter distribution is uniform and has no clutter edges, the assumption is invalid, and all samples x are selected and retained; when At that time, the clutter on both sides differed significantly, and the clutter distribution had clutter edges. The estimated location of the clutter edges was calculated. ; Based on clutter edge location estimation Determine the clutter edge location and select the retained one-sided sample set u, when The clutter edge appears to the right of sample x, taking u=y; when The clutter edge appears to the left of sample x, so we take u=z; Among them, the clutter edge location estimate The total log-likelihood value is calculated as follows:

[0048] in, This represents the total log-likelihood value.

[0049] Specifically, due to the reference window length 2 N Shorter, hypothesis test statistic T C obey t Distribution, marginal test statistic T C critical value T crit significance level of the test Correlation, obtained through the calculation of the marginal test statistic T C Corresponding critical value T crit You can use a lookup table to find... t The distribution table is obtained.

[0050] In some possible implementations of the first aspect, the step involves calculating the sample mean and standard deviation using the Gumbel distribution parameters to obtain Gumbel distribution parameter estimates; calculating weights using the distribution parameter estimates and the standardized residuals defined by the definition to obtain weighting coefficients; constructing a negative likelihood estimation function using the retained one-sided samples and the Gumbel distribution parameters; constructing an objective function based on the weighting coefficients and the negative likelihood estimation function; and solving for the outlier vector, specifically including: The scale parameter of the Gumbel distribution is obtained from the Gumbel distribution. and position parameters The estimated parameters of the Gumbel distribution are obtained by calculating the sample mean and standard deviation. and By defining each sample relative to the position parameter Standardized residuals Specifically:

[0051] Based on standardized residuals Calculate the weights to obtain the weighting coefficients. Specifically:

[0052] The negative likelihood estimation function for one-sided samples is obtained through calculation. Specifically:

[0053] in, K It is the number of samples in the one-sided sample set u; The objective function for outlier detection is constructed based on the negative likelihood estimation function, specifically as follows:

[0054] in, For outlier vectors, , For outlier parameters, The outlier is defined as follows: .

[0055] In some possible implementations of the first aspect, the construction of an outlier index function using outlier vectors, and the selection of outliers based on the values ​​of the outlier vectors according to the outlier index function to obtain the zero norm of the index function, specifically includes: Solving for outlier vectors by minimizing the objective function yields the optimization problem for non-convex outlier vectors:

[0056] Using Taylor's formula for single variables Construct an upper bound function for each single variable. The optimal solution is:

[0057] Among them, operators This operator ensures , k It is the number of iterations; when When, the parameters are obtained. The closed-form solution is as follows:

[0058] in, This indicates the probability of classifying a normal data sample as an outlier. Given all relevant parameters, solve for the outlier vector o. Based on the value of outlier vector o, perform outlier filtering to obtain the outlier index function and the zero norm of the index function, specifically:

[0059] in, The outlier index function is represented by the index function. This represents the zero norm of the index function. K It is the number of samples in a one-sided sample set u.

[0060] In some possible ways of implementing the first aspect, such as Figure 4 As shown, the step of calculating the background clutter power estimate of the current sample using the outlier vector index function and the zero norm of the index function to obtain the corresponding detection threshold, and then determining the target based on the detection threshold, specifically includes: Background clutter power estimates are calculated based on the outlier vector index function and the zero norm of the index function. Specifically:

[0061] Estimated background clutter power Multiplying the result by a preset threshold coefficient yields the corresponding detection threshold, specifically:

[0062] in, T Represents the threshold factor; By comparing the target amplitude D of the unit to be detected with the corresponding detection threshold S, if If the point is determined to be the target point, then the target point is determined to be the target point; otherwise, the target does not exist.

[0063] The ground target detection method proposed in this application is described in detail with reference to embodiments: The radar echo data is down-converted, decimated, compressed, and coherently accumulated to form range-Doppler two-dimensional data. The processed range-Doppler two-dimensional data is used as the original dataset x for target detection. Assume the total length of the reference window is 2N=32.

[0064] like Figure 2 As shown, there is at most one clutter edge within the preset reference window. It is assumed that the clutter edge appears at the [missing information - likely a specific location or event]. M The first reference unit and the first M Between +1 reference units, take sequentially M =1,2…31, get all M The values ​​of the sample sets on both sides, M The values ​​of the left sample set y and M The value of the sample set z to the right; The log-likelihood function is constructed based on the probability density function of the Weibull distribution. Newton's method is used to iteratively calculate the parameter estimates of the sample sets y and z. The specific objective function is as follows:

[0065] Calculate clutter edge statistics TC and critical value T crit Compare and judge in turn M Are the points =1, 2…31 clutter edges? Edge statistics. T C Specifically:

[0066] in, and These are the estimated variances of the scaling parameters for the sample sets y and z, respectively. If all M If none of the samples are determined to be clutter edges, the current reference window is considered to have no clutter edges. The entire original sample set is retained, u=x is set, and the detection ends. If clutter edges exist... M The area is identified as a clutter edge, and the total log-likelihood value is constructed to calculate the estimated location of the clutter edge. Among them, the total log-likelihood value And clutter edge location estimates Specifically:

[0067] Based on clutter edge location estimation Dynamically select a reference window on one side: like This indicates that the clutter edge appears in the right half of the original sample set x, so we take u=y; like This indicates that the clutter edge appears in the left half of the original sample set x, so we take u=z.

[0068] like Figure 3 As shown, the original sample distribution is transformed into a Gumbel distribution, and the estimated values ​​of the Gumbel distribution parameters are calculated. and ,in The default value is 0.5. The estimation is performed using the sample mean and standard deviation, specifically as follows:

[0069] in, The sample mean. The standard deviation of the sample is 1. Here is Euler's constant; Calculate the weighting coefficients based on the estimated distribution parameters. Based on weighted coefficients Calculate outlier parameters Specifically:

[0070] in, This represents the probability of classifying a normal data sample as an outlier, which is the preset false alarm probability. Based on the negative likelihood estimation function of the sample Construct the objective function Specifically:

[0071] in, For outlier vectors, , For outlier parameters, The outlier part is specifically:

[0072] Based on minimizing the objective function Obtain the outlier vector The analytical expression is as follows:

[0073] Since the analytical expression of the outlier vector is a non-convex function, the MM algorithm is used to iteratively solve for the outlier vector. Specifically:

[0074] Based on outlier vectors Constructing the outlier vector index function and index function zero norm I Specifically:

[0075] like Figure 4 As shown, based on the outlier vector index function and index function zero norm I Calculate the background clutter power estimate Specifically:

[0076] Background clutter power estimate With preset threshold factor T Multiply to obtain the detection threshold S Specifically:

[0077] Compare the target amplitude of the unit under test.D With detection threshold S The size, if Then this point is considered the target point. If the target does not exist, then it is determined that the target does not exist.

[0078] Simulation methods were used to verify the performance of this application in operation, such as... Figure 5-8 As shown, the background clutter is assumed to be Weibull distribution clutter, with parameters... The total number of distance units is set to 400, the number of protection units is 2, and the number of left and right reference samples is [missing information]. The number of samples in the CMLD-CFAR (Pruned Mean Constant False Alarm Rate) algorithm Sample number of OS-CFAR algorithm The cutoff factor for TS-CFAR (Truncation Statistical Constant False Alarm Rate) is 2.5, and the average power of clutter in range cells 1-200 is... =22dB, average power of clutter from level 201 to 400 =42dB. Simulations were performed for two scenarios: the target is located far from the clutter edge and the target is located close to the clutter edge. The target was set to be at the 170th and 190th range cells, respectively. The target signal-to-clutter ratio (SCR) was set to 20dB, and the false alarm probability was set to 42dB. Magnitude. Figure 5 and Figure 6 These are the target detection thresholds for two scenarios: the target is located close to the clutter edge, and the target is located far from the clutter edge. Figure 7 This is a comparison of the detection probability of this application with the detection probabilities of six other detection algorithms. Figure 8 This paper compares the false alarm probabilities of this application with those of six detection algorithms when no target is found. Figure 7 and Figure 8 The six detection algorithms described are: CA-CFAR (Cell Average Constant False Alarm Rate), GO-CFAR (Goal-Option-Maximum Constant False Alarm Rate), SO-CFAR (Goal-Option-Minimum Constant False Alarm Rate), CMLD-CFAR (Cut-Off Mean Constant False Alarm Rate), OS-CFAR (Ordered Statistical Constant False Alarm Rate), and TS-CFAR (Truncation-Off Statistical Constant False Alarm Rate). Combined with... Figure 7 and Figure 8 It is evident that, compared with the six detection algorithms, the detection threshold of this application is not affected by strong clutter and clutter edges. The threshold is stable and the step position is close to the clutter edge. This application has better detection performance at clutter edges, good false alarm maintenance capability in weak clutter regions, and improves the false alarm spike phenomenon in strong clutter regions. Therefore, the ground target detection method proposed in this application has good false alarm stability and detection performance in clutter edge environments.

[0079] Secondly, this application provides a ground target detection system with enhanced clutter edge performance, the target detection system including a module for implementing the aforementioned ground target detection method with enhanced clutter edge performance.

[0080] Thirdly, this application provides an electronic device, which can be any device capable of implementing the aforementioned ground target detection method that enhances clutter edge performance. The device can be various terminal devices, such as desktop computers, laptops, tablets, handheld devices, etc., and can be implemented through software and / or hardware.

[0081] For example, the device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is stored in the memory and configured to be executed by the processor to implement the ground target detection method with enhanced clutter edge performance.

[0082] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as C, VHDL, Verilog, the object-oriented programming language Java, and the interpreted scripting language JavaScript.

[0083] This application is described with reference to flowchart illustrations of methods, apparatus, and computer program products according to embodiments of this application. It should be understood that each step in the flowchart can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart.

[0084] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 The function specified in one or more processes.

[0085] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 A process or multiple process diagrams specify the steps of a function.

[0086] In the description of this application, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0087] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0088] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A ground target detection method with enhanced clutter edge performance, characterized in that, include: There is at most one clutter edge within the reference window of the preset radar echo data sample. The clutter edge position is determined by the clutter edge position estimate, and the retained sample set is selected based on the clutter edge position. By combining the estimated distribution parameters with the weighting coefficients obtained from the standardized residuals, the objective function is constructed using the negative likelihood estimation function of the sample. The outlier vector is then obtained by solving the outlier vector index function constructed from the outlier vector. Outlier values ​​are then screened based on the outlier vector index function to obtain the zero norm of the index function. The background clutter power estimate of the current sample is calculated by using the outlier vector index function and the zero norm of the index function, and the corresponding detection threshold is obtained. Target determination is then performed based on the detection threshold.

2. The ground target detection method with enhanced clutter edge performance according to claim 1, characterized in that, The process of determining the clutter edge location using the estimated clutter edge location and selecting a retained sample set based on the clutter edge location specifically includes: Assuming that clutter edges exist in the samples within the reference window, the log-likelihood function is constructed based on the probability density function of the Weibull distribution using the sample data on both sides of the clutter edge, and the estimated clutter parameters are calculated. Calculate the margin test statistic based on the obtained clutter parameter estimates; The existence of clutter edges is determined by the obtained edge test statistics. If no clutter edges exist, the hypothesis is not valid, and the entire sample set is selected and retained. If clutter edges exist, the estimated location of the clutter edges is calculated. The clutter edge location is determined based on the obtained clutter edge location estimate, and a retained one-sided sample set is selected.

3. The ground target detection method with enhanced clutter edge performance according to claim 2, characterized in that, The weighted coefficients obtained by combining the estimated distribution parameters and standardized residuals are used to construct the objective function using the negative likelihood estimation function of the samples. The outlier vector is then solved, and outlier values ​​are screened based on the outlier vector index function constructed from the outlier vectors to obtain the zero norm of the index function. Specifically, this includes: Transform the samples within the reference window into a Gumbel distribution, and use the Gumbel distribution parameters to calculate the sample mean and standard deviation to obtain the estimated Gumbel distribution parameters. The weights are calculated using the estimated distribution parameters and the standardized residuals obtained by definition, resulting in the weighting coefficients. Construct a negative likelihood estimation function using the retained one-sided samples and the parameters of the Gumbel distribution; The objective function is constructed based on the weighting coefficients and the negative likelihood estimation function, and the outlier vector is obtained by solving it. An outlier index function is constructed using outlier vectors. Based on the outlier index function, outlier values ​​are filtered according to the values ​​of the outlier vectors to obtain the zero norm of the index function.

4. The ground target detection method with enhanced clutter edge performance according to claim 2, characterized in that, The probability density function based on the Weibull distribution constructs a log-likelihood function using sample data from both sides of the clutter edge to calculate estimated clutter parameters. The edge test statistic is then calculated based on the obtained clutter parameter estimates, specifically including: The total length of the sample within the preset reference window is 2. N Assuming the clutter edge appears in the first... M The first reference unit and the first M Between +1 reference units, where M The range of values ​​is , to obtain the preceding M The sample set y of each sample and the sample set after inclusion The sample set z of samples; Based on the probability density function of the Weibull distribution, a log-likelihood function is constructed using the two sample sets y and z. Log-likelihood estimation is then performed on the unknown parameters. By solving the nonlinear equation of the log-likelihood estimation, the estimated scale parameter of the sample is obtained. and Shape parameter estimates and ; Based on the obtained scale parameter estimates and Calculate the marginal test statistic T C : in, and These are the estimated variances of the sample parameters on both sides.

5. The ground target detection method with enhanced clutter edge performance according to claim 4, characterized in that, The process involves determining the existence of clutter edges using the obtained edge test statistics. If no clutter edges exist, the hypothesis is invalid, and the entire sample set is retained. If clutter edges exist, an estimated clutter edge location is calculated. The clutter edge location is determined based on the estimated location, and the retained one-sided sample set is selected. This process specifically includes: The obtained margin test statistic T C The critical value of the pre-set test statistic T crit Compared to determining whether clutter edges exist, when When the clutter distribution is uniform and has no clutter edges, the assumption is invalid, and all samples x are selected and retained; when At that time, the clutter on both sides differed significantly, and the clutter distribution had clutter edges. The estimated location of the clutter edges was calculated. ; Based on clutter edge location estimation Determine the clutter edge location and select the retained one-sided sample set u, when The clutter edge appears to the right of sample x, taking u=y; when The clutter edge appears to the left of sample x, so we take u=z; Among them, the clutter edge location estimate The total log-likelihood value is calculated as follows: in, This represents the total log-likelihood value.

6. The ground target detection method with enhanced clutter edge performance according to claim 5, characterized in that, The sample mean and standard deviation are calculated using the Gumbel distribution parameters to obtain estimated Gumbel distribution parameters; weights are calculated using the estimated distribution parameters and the standardized residuals defined by the definition to obtain weighting coefficients; Construct a negative likelihood estimation function using the retained one-sided samples and the parameters of the Gumbel distribution; The objective function is constructed based on the weighted coefficients and the negative likelihood estimation function. The outlier vector is obtained by solving the objective function, specifically including: The scale parameter of the Gumbel distribution is obtained from the Gumbel distribution. and position parameters The estimated parameters of the Gumbel distribution are obtained by calculating the sample mean and standard deviation. and By defining each sample relative to the position parameter Standardized residuals Specifically: Based on standardized residuals Calculate the weights to obtain the weighting coefficients. Specifically: The negative likelihood estimation function for one-sided samples is obtained through calculation. Specifically: in, K It is the number of samples in the one-sided sample set u; The objective function for outlier detection is constructed based on the negative likelihood estimation function, specifically as follows: in, For outlier vectors, , For outlier parameters, The outlier is defined as follows: 。 7. The ground target detection method with enhanced clutter edge performance according to claim 6, characterized in that, The process of constructing an outlier index function using outlier vectors, and then filtering outliers based on the values ​​of the outlier vectors to obtain the zero norm of the index function, specifically includes: Solving for outlier vectors by minimizing the objective function yields the optimization problem for non-convex outlier vectors: Using Taylor's formula for single variables Construct an upper bound function for each single variable. The optimal solution is: Among them, operators This operator ensures , k It is the number of iterations; when When, the parameters are obtained. The closed-form solution is as follows: in, This indicates the probability of classifying a normal data sample as an outlier. Given all relevant parameters, solve for the outlier vector o. Based on the value of outlier vector o, perform outlier filtering to obtain the outlier index function and the zero norm of the index function, specifically: in, The outlier index function is represented by the index function. This represents the zero norm of the index function. K It is the number of samples in a one-sided sample set u.

8. The ground target detection method with enhanced clutter edge performance according to claim 7, characterized in that, The process of calculating the background clutter power estimate of the current sample using the outlier vector index function and the zero norm of the index function to obtain the corresponding detection threshold, and then determining the target based on the detection threshold, specifically includes: Background clutter power estimates are calculated based on the outlier vector index function and the zero norm of the index function. Specifically: Estimated background clutter power Multiplying the result by a preset threshold coefficient yields the corresponding detection threshold, specifically: in, T Represents the threshold factor; By comparing the target amplitude D of the unit to be detected with the corresponding detection threshold S, if If the point is determined to be the target point, then the target point is determined to be the target point; otherwise, the target does not exist.

9. A ground target detection system with enhanced clutter edge performance, characterized in that: Includes modules for implementing the ground target detection method with enhanced clutter edge performance as described in any one of claims 1-8.

10. An electronic device, characterized in that: The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the ground target detection method with enhanced clutter edge performance as described in any one of claims 1-8.

Images