An information geometry multi-frame nonlinear detection method for unknown extended point target in non-uniform clutter environment

Abstract

The application discloses an information geometry multi-frame nonlinear detection method for unknown extended point number targets in a non-uniform clutter environment, and belongs to the technical field of radar target detection. The method comprises the following steps: constructing joint measurement data on the basis of echoes, mapping the joint measurement data to an extended target manifold space, and obtaining an extended target joint measurement manifold coordinate; then, estimating a multi-frame clutter geometric center by using historical data of several frames before a specified moment, so as to suppress the influence of non-uniform clutter variation on a subsequent target detection process; finally, accumulating a nonlinear value function of the extended target, and outputting an estimated result of an unknown extended point number target track and an extended point number. The application solves the problem that, in a non-uniform clutter environment, the estimated clutter geometric center deviates from the clutter distribution characteristics of a unit to be detected due to the rapid variation of the clutter statistical characteristics with the spatial position, thereby leading to the decline of the detection performance, and solves the problem of the loss of the manifold measurement of a weak extended target, and improves the detection performance of the weak extended target in the non-uniform clutter environment.

Description

Technical Field

[0001] This application belongs to the field of radar target detection technology, and in particular relates to an extended target multi-frame detection method based on information geometry in non-uniform clutter environments. Background Technology

[0002] In recent years, the research and application of information geometry methods in radar detection have developed rapidly, becoming one of the important technical means in complex clutter environments. This method maintains excellent performance even with limited clutter samples, does not rely on prior knowledge of clutter distribution, and possesses significant clutter suppression capabilities, thus effectively improving the stability and reliability of target detection. Multi-frame detection can significantly enhance weak target detection capabilities by integrating information from the temporal dimension, but its performance deteriorates severely in clutter backgrounds. To address this issue, combining information geometry with multi-frame detection can improve the accuracy of clutter modeling while effectively aggregating weak target features, thereby significantly enhancing target detection performance in complex scenes.

[0003] In existing radar information geometry detection methods, such as "A Method for Detecting Weak Targets in Subband Information Geometry Radar Based on Orthogonal Projection" by Yang Zheng et al. and the Chinese invention patent application CN202211598623.4, "A Method and Device for Detecting Power Spectrum Information Geometry Radar Targets Based on Subband Filtering," the echo sample of a single range cell is typically modeled as a coordinate point on a manifold for subsequent geometric distance calculation and decision-making. However, with the continuous improvement of radar resolution, in practical applications such as airborne and vehicle-mounted radar, the physical size of many targets often spans multiple range resolution cells, thus exhibiting obvious range extension characteristics. Most existing methods are based on point target models, typically assuming that the echo samples of each range cell are independent. However, for extended targets, their echoes exhibit significant correlations across multiple range cells. If the aforementioned independence assumption is still adopted, it is difficult to effectively characterize the spatial distribution of scattering centers, resulting in insufficient modeling of the overall target structure. Furthermore, existing information geometry methods can only describe the local features of extended targets, failing to fully reflect their overall characteristics, thus further limiting the improvement of extended target detection performance. Although existing literature has studied extended target information geometry, none of it has considered non-uniform clutter environments where clutter statistics change rapidly with spatial location. Information geometry detection algorithms rely on accurate estimation of clutter distribution. In non-uniform clutter environments, the rapid changes in clutter statistics with spatial location make it difficult for existing clutter estimation methods to accurately characterize the true clutter statistics of the cell undertest (CUT), resulting in a significant decrease in detection performance. Summary of the Invention

[0004] The purpose of this application is to overcome the shortcomings of the prior art and provide an information geometry multi-frame nonlinear detection method for targets with unknown extended point numbers in non-uniform clutter environments. This method addresses the problem that the limited reference samples used to characterize the true clutter distribution type of the CUT in non-uniform clutter environments lead to clutter modeling deviations, thereby limiting detection performance, and the problem of manifold metric loss in information geometry detection algorithms for extended targets.

[0005] The objective of this application is achieved through the following technical solution: A geometric multi-frame nonlinear detection method for targets with unknown spread points in a non-uniform clutter environment, the method comprising: The extended target is subjected to multiple hypothesis processing. Joint measurement data is constructed based on the measurement values ​​of multiple scattering units. Then, the joint measurement data is mapped onto the manifold to obtain the joint metric manifold coordinates of the extended target. Using historical data from several frames prior to a specified time, estimate the geometric center of clutter in multiple frames; Based on the geometric center of the multi-frame clutter, the nonlinear value function accumulation function of the extended target is substituted to perform an accumulation decision, and the estimated values ​​of the target trajectory and the number of target extension points are output.

[0006] Furthermore, the process of performing multiple hypothesis processing on the extended target, constructing joint measurement data based on the measurements of multiple scattering units, and then mapping the joint measurement data onto the manifold to obtain the joint metric manifold coordinates of the extended target, specifically includes: The number of expansion points of the expanded target is represented as a multi-hypothesis testing problem. All hypotheses are traversed, and under each hypothesis, the corresponding joint measurement data is obtained. Then, the joint metric manifold coordinates of the expanded target under that hypothesis are obtained.

[0007] Furthermore, by utilizing historical data from several frames prior to a specified time, the estimation of the multi-frame clutter geometric center specifically includes: The target cell in each frame is determined, and reference cells around the target cell are selected to obtain a training sample set for estimating clutter parameters.

[0008] Furthermore, the method also includes: In non-uniform clutter environments, joint measurement data contained in adjacent frame reference windows are also used as training samples to estimate clutter parameters.

[0009] Furthermore, the method obtains the threshold for the decision corresponding to the specified false alarm rate through multiple Monte Carlo experiments.

[0010] Furthermore, the method also includes obtaining the target estimated state through track backtracking.

[0011] The beneficial effects of this application are as follows: The method of the present invention uses joint measurement data to obtain the joint metric manifold coordinates of the extended target, accumulates the features of the extended target on the manifold, and realizes the information collection of the extended target.

[0012] This invention utilizes a multi-frame clutter geometric center joint estimation method to improve the accuracy of clutter estimation in non-uniform clutter environments. At the same time, multi-frame accumulation of geometric distance enhances the ability to identify weak targets. Compared with existing information geometry algorithms, it can more efficiently identify extended targets. Attached Figure Description

[0013] Figure 1 This is a flowchart of the information geometric multi-frame nonlinear detection method for targets with unknown spread points in non-uniform clutter environments provided by the present invention.

[0014] Figure 2 This is a schematic diagram of the radar echo of the present invention.

[0015] Figure 3 shows the processing results of various algorithms for measured radar data in the ground clutter scene. (a) is the detection statistics plane of MTD in the last frame; (b) is the detection statistics plane of extended target energy weighted multi-frame detection; (c) is the detection statistics plane of matrix information geometric multi-frame detection; (d) is the detection statistics plane of power spectrum information geometric multi-frame detection; (e) is the detection statistics plane of extended target matrix information geometric multi-frame detection; and (f) is the detection statistics plane of extended target power spectrum information geometry.

[0016] Figure 4 The results are multi-frame detection results with the measured sea clutter in the IPIX dataset as the background. (a) is the detection probability curve and (b) is the root mean square error curve. Detailed Implementation

[0017] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0018] Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] In existing radar information geometric detection methods, the limited reference samples used to characterize the actual clutter distribution type of the CUT lead to clutter modeling bias and difficulty in effectively collecting extended target features due to model mismatch in extended target scenarios. This results in the geometric distance failing to reflect the actual difference between the target and clutter, thus limiting detection performance.

[0020] To address the aforementioned technical problems, the following embodiments of the information geometric multi-frame nonlinear detection method for targets with unknown spread point counts in non-uniform clutter environments are proposed.

[0021] This embodiment provides an information geometry multi-frame nonlinear detection method for targets with unknown extended points in a non-uniform clutter environment. The method first integrates range cell features using multiple hypothesis windows to collect extended target features and obtain joint measurement data. The joint measurement data is then mapped to a positive definite Hermitian (HPD) manifold space or a power spectrum manifold space to obtain the joint metric manifold coordinates of the extended targets. Next, multi-frame clutter samples are used for joint estimation to obtain the multi-frame clutter geometric center. Then, the geometric distance between the joint metric manifold coordinates of the extended targets of the target cell and the multi-frame clutter geometric center is calculated to quantify the overall difference between the target cell and the internal structure of the clutter to obtain a nonlinear value function. Finally, the nonlinear value function is accumulated over multiple frames and a decision is made.

[0022] Reference Figure 1 ,like Figure 1 The diagram shown is a flowchart of a geometric multi-frame nonlinear detection method for targets with unknown spread points in a non-uniform clutter environment provided in this embodiment. The method includes the following steps: Step 1: Extend the joint metric manifold coordinate mapping of the target.

[0023] Targets move relatively slowly within clutter regions, and multi-frame detection typically involves batch processing of echo data from short intervals; therefore, the number of scattering points for extended targets usually remains almost constant between these frames. (Refer to...) Figure 2 ,like Figure 2 The diagram shown is a schematic of the radar echo in this embodiment.

[0024] For radar echoes from rigid extended targets a A series of strong scattering points, centered at their centroids, are denoted as... t The state of the center of mass at time t is The corresponding radar measurement vector is composed of radial distance. and speed The composition can be represented as: ; in, yes tMeasurement noise at any given time. Assume... a The velocities of the consecutive scattering points are the same and equal to... The set of scattering points of the extended target can be represented as: ; in, This represents the distance resolution.

[0025] In reality, the number of expansion points for the expanded objective is unknown, therefore it can be represented as a multiple hypothesis testing problem. For the th... t , For frame echoes, one of the following assumptions holds: ; In the assumption Below, using a length of k A sliding window is used to integrate distance cell information from the distance dimension samples to obtain joint measurement data: ; Indicates the new first t Frame, number r Combined measurement data. Iterate through all hypotheses following the steps described above.

[0026] exist Assuming, let's denote... Its covariance matrix is: ; ; in, M This represents the number of pulses emitted within a single pulse repetition interval.

[0027] The corresponding power spectrum is ,Depend on It consists of eigenvalues.

[0028] Step 2: Joint estimation of clutter samples from multiple frames. for t Real-time radar measurements can be used in combination of l Frame history data is used to calculate the geometric center of clutter. Therefore, it is necessary to first determine the cutout for each frame, and then select reference cells around the cutout to calculate the geometric center. Given the extended target centroid at time t, it can be derived from the following formula. t - n The target state at any given moment.

[0029] ; ; in, The mean is The covariance matrix is The Gaussian distribution. F and the above equation R The state transition matrix and the process noise covariance matrix are respectively. .

[0030] for ,exist t - n The reference cell sample at time t can be represented as ; in, To protect window size, This is for reference window size. It serves as a single-frame reference window.

[0031] For non-uniform clutter environments, the corresponding range cells within the reference window of adjacent frames are also used as reference cells for estimating clutter parameters. t The reference unit training sample set at time step 1 can be represented as ; The geometric center of the new sample on the matrix manifold after integrating the distance cell information is: ; in, This represents the total number of reference units.

[0032] The geometric center of the new sample after integrating the distance cell information is on the power spectral manifold of the reference cell. ; Step 3: Expand the accumulation of the target nonlinear value function. Through multiple Monte Carlo experiments, the threshold corresponding to a certain false alarm rate can be obtained. The extended target matrix information geometry is accumulated over multiple frames using the following formula: ; ; in, In order to be in Assume the possible state range of the target centroid, and let E be the identity matrix. The single-frame detection statistics plane is obtained through the above formula.

[0033] For the extended target power spectral information geometry, multi-frame accumulation is achieved through the following formula: ; ; in, and They are respectively and The i Each element.

[0034] Step 4: Value function decision and target parameter output. The backtracking expression is .

[0035] Accumulation T Value function after frame data After threshold decision, the final estimated target state sequence can be obtained by the following formula: ; ; .

[0036] The previous path was obtained by tracing back the flight path. T -1 frame target estimation state The estimated state of the target at time step can be expressed as: .

[0037] This embodiment's method, based on an information geometry framework, effectively characterizes the statistical properties of clutter in non-uniform clutter environments by jointly estimating multiple frames of clutter samples, thereby improving the accuracy of clutter parameter estimation. Building upon this, it utilizes a joint metric manifold coordinate mapping for extended targets to accumulate extended target features in the manifold space. Finally, it performs multi-frame accumulation detection using a nonlinear value function, effectively enhancing weak extended target information. Compared to existing information geometry extended target detection multi-frame algorithms, this method achieves more robust extended target identification in non-uniform clutter environments.

[0038] This embodiment also uses simulation experiments on Matlab R2024b for verification. The specific verification process is as follows: Step 1: Using the sea clutter in 19980205_184403_antstep.cdf as the clutter background, an extended target with a range cell extended is injected. The radar transmits 10 pulses within one pulse repetition interval. The raw radar echo is filtered using a filter bank containing 10 128th-order Caesar windows to obtain... Referring to Figure 3, which shows the detection statistics planes of different methods for measured extended target data in a ground clutter scenario, (a) is the detection statistics plane of the last frame's MTD; (b) is the detection statistics plane of extended target energy-weighted based multi-frame detection (ET-EW-MFD); (c) is the detection statistics plane of matrix information geometry multi-frame detection (MIG-MFD); (d) is the detection statistics plane of power spectrum information geometry multi-frame detection (PSIG-MFD); (e) is the detection statistics plane of one of the methods in this paper, namely extended target matrix information geometry multi-frame detection (ET-MIG-MFD); and (f) is the detection statistics plane of one of the methods in this paper, namely extended target power spectrum information geometry multi-frame detection (ET-PSIG-MFD). This embodiment uses measured data of extended targets in a ground clutter scenario as an example.

[0039] Step 2: Integrate the echo data from Step 1, assuming... , , and Then, the integrated joint measurement data is mapped to the manifold space.

[0040] Step 3: Estimate clutter parameters using clutter samples from the previous 3 frames and the current frame. Set the number of guard elements to 1 and the number of reference elements to 40 to obtain the geometric center of multiple frames.

[0041] Step 4: Using the results obtained in the previous steps and or and Calculate the geometric distance to obtain the nonlinear value function.

[0042] Step 5: Accumulate the nonlinear value function obtained in Step 4 over multiple frames, accumulating 4 frames of data at a time to obtain the detection statistics planes for different algorithms.

[0043] Step Six: Using the IPIX dataset 19980205_184403_antstep.cdf as the clutter background, repeat the above steps. Perform 5000 Monte Carlo experiments with pure clutter based on the first 50030 pulses to obtain the threshold under different assumptions, thus controlling the actual false alarm rate. Then, 100 Monte Carlo experiments were performed on the remaining pulse injection targets to obtain the corresponding detection probability curve and root mean square error curve. (Refer to...) Figure 4 ,like Figure 4 The results shown are multi-frame detection results against the background of measured sea clutter in the IPIX dataset.

[0044] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A multi-frame nonlinear detection method for targets with unknown spread points in a non-uniform clutter environment, characterized in that, The method includes: The extended target is subjected to multiple hypothesis processing, and joint measurement data is constructed based on the measurement values ​​of multiple scattering units. Then, the joint metric manifold coordinates of the extended target are obtained. Using historical data from several frames prior to a specified time, estimate the geometric center of clutter in multiple frames; Based on the geometric center of the multi-frame clutter, the nonlinear value function accumulation function of the extended target is substituted into the accumulation decision, and the target track and the number of extension points are output.

2. The information geometric multi-frame nonlinear detection method for targets with unknown spread points in a non-uniform clutter environment as described in claim 1, characterized in that, The process of performing multiple hypothesis processing on the extended target to obtain the joint metric manifold coordinates of the extended target specifically includes: The number of extended points of the extended target is represented as a multi-hypothesis testing problem. All hypotheses are traversed, and under each hypothesis, the radar echoes are constructed into corresponding joint measurement data, which are then mapped onto the manifold to obtain the joint metric manifold coordinates of the extended target.

3. The information geometric multi-frame nonlinear detection method for targets with unknown spread points in a non-uniform clutter environment as described in claim 1, characterized in that, The step of calculating the clutter geometric center at the specified time using historical data from several frames prior to the specified time specifically includes: The cell to be detected in each frame is determined, and reference cells around the cell to be detected are selected for calculating the geometric center.

4. The information geometric multi-frame nonlinear detection method for targets with unknown spread points in a non-uniform clutter environment as described in claim 3, characterized in that, The method further includes: In non-uniform clutter environments, joint measurement data contained in the reference windows of previous frames and several historical frames are also used as training samples to estimate clutter parameters.

5. The information geometric multi-frame nonlinear detection method for targets with unknown spread points in a non-uniform clutter environment as described in claim 1, characterized in that, The method obtains the threshold for decision-making at a specified false alarm rate through multiple Monte Carlo experiments.

6. The information geometric multi-frame nonlinear detection method for targets with unknown spread points in a non-uniform clutter environment as described in claim 1, characterized in that, The method also includes obtaining the estimated target state and the estimated number of unknown extended target points through track backtracking and extended target structure parameter estimation.

Images