An Extended Target Adaptive Detection Method and System for Strong Clutter Environments

Abstract

This invention discloses an adaptive detection method and system for extended targets in strong clutter environments. It addresses the performance degradation of traditional point target detection methods caused by extended targets spanning multiple range cells and the potential obstruction of the signal subspace by the clutter subspace in strong clutter environments. The method of this invention determines the target data and training sample model based on environmental and radar system characteristics, and establishes a binary hypothesis testing model; it then establishes a joint probability density function based on the hypothesis testing model; using the joint probability density function, it estimates unknown parameters under the GLRT and Rao criteria respectively, constructing detection statistics; it determines a detection threshold based on the detection statistics and a preset false alarm probability; and it compares the detection statistics with the threshold to determine the presence of the target. This method effectively improves the detection performance of extended targets in strong clutter environments, is applicable to the adaptive detection of extended targets such as ships, aircraft, and ground targets, and possesses constant false alarm characteristics and good engineering application value.

Description

Technical Field

[0001] This invention belongs to the field of radar signal processing and adaptive detection technology, specifically relating to an extended target adaptive detection method and system in a strong clutter environment. Background Technology

[0002] In modern radar detection systems, the sea surface, land surface, and complex electromagnetic environments are often accompanied by strong clutter backgrounds, such as sea clutter, land clutter, and multipath reflection clutter. These clutters often exhibit significant structural characteristics and statistical correlations, and their power levels can be far higher than the target echo, severely impacting the detection performance of the radar system. Under strong clutter conditions, traditional detection methods based on simple thresholds or independent identical distribution assumptions struggle to effectively distinguish between target and clutter components, especially under low signal-to-noise ratio conditions, which can easily lead to missed detections or false alarms.

[0003] Meanwhile, real-world targets often exhibit extended target forms, with their echoes typically spanning multiple range cells or observation snapshots, containing multiple scattering centers, and displaying extended spatial and energy distribution characteristics. Compared to point target models, extended targets require joint modeling and statistical processing of multiple observation cells; otherwise, it is difficult to fully utilize the target echo energy. In strong clutter environments, extended target echoes often superimpose with clutter components or even structurally couple, making it difficult to directly apply traditional adaptive point target detection algorithms.

[0004] Furthermore, in strong clutter backgrounds, clutter components can typically be represented as random processes located in a high-dimensional subspace, while the target signal subspace may be embedded in or confined within this clutter subspace. This leads to a significant performance degradation in detection methods based on subspace orthogonal separation or simple projection suppression. Existing extended target detection techniques still suffer from problems such as insufficient modeling, inadequate adaptive estimation accuracy, and difficulty in guaranteeing constant false alarm rate characteristics under such subspace-constrained conditions.

[0005] A search revealed that existing patent CN120871058A discloses a radar extended target detection method and device under composite Gaussian clutter. Its core lies in constructing a training sample matrix, a signal matrix, and a target data matrix, but it does not specifically address scenarios where the target subspace is constrained by the clutter subspace. CN120275924A, a fully adaptive constant false alarm rate (CFAR) detection method for generalized clutter texture distribution, focuses on texture distribution modeling and Monte Carlo offline threshold estimation, which is substantially different from the subspace constraint modeling method of this invention. Furthermore, existing technologies mostly choose between the GLRT criterion and the Rao criterion to design detectors, lacking a flexible approach that simultaneously provides detection statistics for both criteria within the same framework.

[0006] Therefore, it is necessary to propose an adaptive detection method and system for extended targets in strong clutter environments to achieve robust detection of extended targets under subspace-constrained conditions. Summary of the Invention

[0007] To address the aforementioned technical challenges, this invention provides an adaptive detection method and system for extended targets in strong clutter environments. This system solves the problem of insufficient extended target detection performance in the prior art under strong clutter backgrounds, thereby improving the detection performance of extended targets in complex electromagnetic environments.

[0008] This invention provides an extended target adaptive detection method for environments with strong clutter, comprising the following steps:

[0009] Step 1: Determine the data model for the data to be detected and the training samples based on environmental and system characteristics;

[0010] Step 2: Establish a binary hypothesis testing model using the data to be tested and the training samples;

[0011] Step 3: Establish the joint probability density function based on the hypothesis testing model;

[0012] Step 4: Using the joint probability density function, estimate the unknown parameters under the GLRT criterion and the Rao criterion respectively, construct the detection statistics, and obtain the detection statistics based on the GLRT criterion and the detection statistics based on the Rao criterion;

[0013] Step 5: Determine the detection threshold based on the detection statistics and the preset false alarm probability;

[0014] Step 6: Compare the detection statistics with the detection threshold and determine whether the target exists.

[0015] This invention also provides an extended target adaptive detection system for environments with strong clutter, comprising:

[0016] The module for constructing detection data and training samples is used to construct detection data and training sample data based on environmental and system characteristics.

[0017] The parameter estimation module is used to estimate unknown parameters under the GLRT criterion and the Rao criterion, respectively.

[0018] The detector construction module is used to construct detection statistics using the GLRT criterion and the Rao criterion, respectively.

[0019] The detection threshold determination module is used to determine the detection threshold using the detection statistics and the preset false alarm probability.

[0020] The target decision module is used to compare the magnitude between the detection statistic and the detection threshold, and to determine whether the target exists.

[0021] Compared with the prior art, the present invention has the following advantages:

[0022] 1. An extended joint target detection model is constructed for strong clutter environments. Compared with traditional point target detection methods, it can effectively suppress the influence of clutter in strong clutter backgrounds and improve the detectability of targets.

[0023] 2. By adopting a multi-distance unit collaborative processing mechanism, the energy of the target across units is fully integrated and expanded, which significantly improves the detection probability and stability compared with the single-unit processing method;

[0024] 3. Parameter estimation and statistical decision-making are achieved under clutter subspace constraints. Compared with existing methods that rely on the subspace independence assumption, this method is more suitable for strong clutter environments where signals and clutter structures are coupled.

[0025] 4. It provides detection statistics under both GLRT and Rao criteria, allowing users to choose the appropriate detector based on their actual needs. The GLRT detector has a higher detection probability, while the Rao detector has better overall performance in certain scenarios, providing flexible options for radar system design. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the extended target adaptive detection method for strong clutter environments according to the present invention;

[0027] Figure 2 This is a system structure framework diagram of the extended target adaptive detection method for strong clutter environments of the present invention;

[0028] Figure 3 This is a schematic diagram comparing the detection probabilities of the method of the present invention under three different criteria and different signal-to-noise ratios. Detailed Implementation

[0029] The implementation of the technical solution of the present invention will be further described below with reference to the embodiments, accompanying drawings, and mathematical models.

[0030] Example 1

[0031] Taking airborne radar arrays as the research object, using 3D column vector This represents the two-dimensional spatiotemporal data to be detected, where, , For the number of array elements, The number of pulses. The number of system channels. To expand the number of distance units occupied by the target.

[0032] An extended target adaptive detection method for environments with strong clutter, referring to... Figure 1 This includes the following steps:

[0033] Step 1: Determine the data model for the data to be detected and the training samples based on environmental and system characteristics;

[0034] In hypothesis testing Below, data to be detected Includes noise components Interference components and signal components Interference components and signal components They are respectively represented as and ;

[0035] in, and They are respectively dimensional target signal subspace and Dimensional interference signal subspace, Let M be an M-row, p-column matrix, both of which are full-rank columns, and the target signal subspace is defined. This is an interference signal. ;

[0036] and These are the corresponding unknown subspace coordinates, with corresponding dimensions of... and ;

[0037] noise components Each column is independently and identically distributed with a mean of 0 and a covariance of . The complex Gaussian distribution;

[0038] Interference component coordinates Each column is independently and identically distributed with a mean of 0 and a covariance of . The complex Gaussian distribution;

[0039] matrix and All parameters are known, matrix and For unknown parameters, the covariance is... and scalar For unknown parameters;

[0040] Suppose there exist L independent and identically distributed training samples, let the i-th... The training samples are , . It contains noise components , It follows a pattern with a mean of 0 and a covariance of . The complex Gaussian distribution also contains interference components. , It follows a pattern with a mean of 0 and a covariance of . The complex Gaussian distribution;

[0041] In hypothesis testing Below, data to be detected Includes noise components and interference components .

[0042] Step 2: Establish a binary hypothesis testing model using the data to be tested and the training samples;

[0043] make A binary hypothesis testing model is established using the data to be detected and the training samples. The binary hypothesis testing problem for extended object detection is expressed as:

[0044] .

[0045] Step 3: Establish the joint probability density function based on the hypothesis testing model as follows:

[0046] ;

[0047] in, , , Indicates taking Exponentiation, Represents matrix trace operation For vector representations, take the inverse operation; for scalar representations, take the reciprocal operation. This represents the summation operation. This represents the matrix conjugate transpose operation. This indicates the matrix determinant operation.

[0048] Step 4: Using the joint probability density function, estimate the unknown parameters under the GLRT and Rao criteria respectively, and construct the detection statistics. The detection statistics based on the GLRT criterion are as follows:

[0049] ;

[0050] The detection statistic based on the Rao criterion is: ;

[0051] in, , .

[0052] Step 5: Based on the obtained detection statistics and preset false alarm probability, determine the detection threshold through theoretical distribution analysis or numerical methods. .

[0053] Step 6: Calculate the detection statistic t and the detection threshold. Compare and determine if the target exists;

[0054] If the detection statistic t is greater than or equal to the detection threshold If the detection statistic t is less than the detection threshold, then the target is determined to exist; If the target does not exist, then it is determined that the target does not exist.

[0055] Example 2

[0056] An extended target adaptive detection system for environments with strong clutter, referring to... Figure 2 It includes modules for constructing data to be detected and training samples, parameter estimation, detector construction, detection threshold determination, and target decision.

[0057] Simulation Experiment

[0058] The effects of the present invention will be further explained below with reference to simulation experiments.

[0059] Figure 3 The results of the detection probability comparison of the detection method of the present invention under different criteria and different signal-to-noise ratios are presented. Figure 3 It can be clearly seen that in a strong clutter environment, the detector based on the GLRT criterion proposed in this invention has the best detection effect, followed by the Wald detector, and finally the Rao detector. The detection probability of all of them gradually increases with the increase of the signal-to-noise ratio.

[0060] In practical applications, either a GLRT detector or a Rao detector can be selected based on system design requirements. Generally, the GLRT detector achieves a high detection probability under all signal-to-noise ratio conditions; the Rao detector has relatively low computational complexity and good robustness to mismatched signals. To accommodate the needs of different application scenarios, the system of this invention can be configured with both detectors simultaneously, allowing users to flexibly choose according to specific scenarios.

[0061] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for extended target adaptive detection in a strong clutter environment, characterized in that, Includes the following steps: Step 1: Determine the data model for the data to be detected and the training samples based on environmental and system characteristics; Step 2: Establish a binary hypothesis testing model using the data to be tested and the training samples; Step 3: Establish the joint probability density function based on the hypothesis testing model; Step 4: Using the joint probability density function, estimate the unknown parameters under the GLRT criterion and the Rao criterion respectively, construct the detection statistics, and obtain the detection statistics based on the GLRT criterion and the detection statistics based on the Rao criterion. Step 5: Determine the detection threshold based on the detection statistics and the preset false alarm probability; Step 6: Compare the detection statistics with the detection threshold and determine whether the target exists.

2. The extended target adaptive detection method for strong clutter environments according to claim 1, characterized in that, In step 1, during hypothesis testing Below, data to be detected Includes noise components Interference components and signal components Interference components and signal components They are respectively represented as and ; in, and They are respectively dimensional target signal subspace and Dimensional interference signal subspace, Let M be an M-row, p-column matrix, both of which are full-rank columns, and the target signal subspace is defined. This is an interference signal. ; and These are the corresponding unknown subspace coordinates, with corresponding dimensions of... and ; noise components Each column is independently and identically distributed with a mean of 0 and a covariance of . The complex Gaussian distribution; Interference component coordinates Each column is independently and identically distributed with a mean of 0 and a covariance of . The complex Gaussian distribution; matrix and All parameters are known, matrix and For unknown parameters, the covariance is... and scalar For unknown parameters; Suppose there exist L independent and identically distributed training samples, let the i-th... The training samples are , . It contains noise components , It follows a pattern with a mean of 0 and a covariance of . The complex Gaussian distribution also contains interference components. , It follows a pattern with a mean of 0 and a covariance of . The complex Gaussian distribution; In hypothesis testing Below, data to be detected Includes noise components and interference components .

3. The extended target adaptive detection method for strong clutter environments according to claim 2, characterized in that, In step 2, let A binary hypothesis testing model is established using the data to be detected and the training samples. The binary hypothesis testing problem for extended object detection is expressed as: 。 4. The extended target adaptive detection method for strong clutter environments according to claim 3, characterized in that, In step 3, the joint probability density function is established based on the hypothesis testing model as follows: ； in, , , Indicates taking Exponentiation, Represents matrix trace operation For vector representations, take the inverse operation; for scalar representations, take the reciprocal operation. This represents the summation operation. This represents the matrix conjugate transpose operation. This indicates the matrix determinant operation.

5. The extended target adaptive detection method for strong clutter environments according to claim 4, characterized in that, In step 4, the unknown parameters are estimated using the joint probability density function under the GLRT and Rao criteria, respectively, and detection statistics are constructed. The detection statistics based on the GLRT criterion are as follows: ； The detection statistic based on the Rao criterion is: ； in, , .

6. The extended target adaptive detection method for strong clutter environments according to claim 5, characterized in that, In step 5, based on the detection statistics and preset false alarm probability obtained in step 4, the detection threshold is determined through theoretical distribution analysis or numerical methods. .

7. The extended target adaptive detection method for strong clutter environments according to claim 6, characterized in that, In step 6, the detection statistic t is compared with the detection threshold. Comparison: If the detection statistic t is greater than or equal to the detection threshold If the detection statistic t is less than the detection threshold, then the target is determined to exist; If the target does not exist, then it is determined that the target does not exist.

8. An extended target adaptive detection system for strong clutter environments, used to implement the method according to any one of claims 1 to 7, characterized in that, Include: The module for constructing detection data and training samples is used to construct detection data and training sample data based on environmental and system characteristics. The parameter estimation module is used to estimate unknown parameters under the GLRT criterion and the Rao criterion, respectively. The detector construction module is used to construct detection statistics using the GLRT criterion and the Rao criterion, respectively. The detection threshold determination module is used to determine the detection threshold using the detection statistics and the preset false alarm probability. The target decision module is used to compare the magnitude between the detection statistic and the detection threshold, and to determine whether the target exists.

Images