Kalman filter-based radar large-target split track condensation method
By using the Kalman filter method to converge split tracks of large radar targets, the problem of inaccurate convergence of split tracks in existing technologies is solved, achieving higher convergence accuracy and tracking stability, and optimizing resource utilization and intelligent management.
Patent Information
- Application Number
- CN202511079507.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-01
- Publication Date
- 2025-11-14
AI Technical Summary
Existing radar track processing algorithms struggle to accurately reassemble split tracks into a form that reflects the true target state when large targets are split, leading to misjudgments and low tracking efficiency.
A Kalman filter-based approach is adopted, which involves initializing the Kalman filter, predicting the track state, associating and matching observation points, verifying the effectiveness of convergence, and performing weighted fusion. Combined with an adaptive Mahalanobis distance threshold and echo feature similarity criteria, a complete track convergence system is constructed.
It improves the accuracy of radar in converging split tracks of large targets, enhances tracking stability, reduces misjudgments, optimizes resource utilization, and supports intelligent management.
Smart Images
Figure CN120949216A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of radar track processing technology, specifically to a radar large target split track convergence method based on Kalman filter. Background Technology
[0002] In radar monitoring, when a large target splits (e.g., due to structural changes or electromagnetic interference), radar may detect multiple seemingly independent tracks, even though these tracks may actually belong to the same large target. Current track processing algorithms are inadequate in handling this situation, failing to accurately reassemble the split tracks into a single track reflecting the true target state. This can lead to misjudgment of the target, low tracking efficiency, and subsequent decision-making errors based on track information.
[0003] Therefore, this invention requires the design of a radar large target splitting trajectory convergence method based on Kalman filter to solve the above-mentioned problems. Summary of the Invention
[0004] The purpose of this invention is to provide a radar large target splitting trajectory convergence method based on Kalman filter to solve the above-mentioned problems, thus solving the problems mentioned in the background art.
[0005] To address the above problems, the present invention provides a technical solution:
[0006] A radar large target split track convergence method based on Kalman filter includes the following specific steps:
[0007] S1. Initialize the points detected by the radar, create an initial track file, and initialize the Kalman filter for each track;
[0008] S2. Predict the state of each track at the next moment using a Kalman filter;
[0009] S3. Associate and match the newly obtained observation points with the predicted state;
[0010] S4. Perform agglomeration judgment on associated tracks that meet the spatial proximity criterion;
[0011] S5. Verify the effectiveness of track convergence based on the motion consistency criterion and echo characteristic similarity criterion;
[0012] S6. Update the Kalman filter state using the validated condensation observations;
[0013] S7, outputs the stable trajectory after condensation.
[0014] In a preferred embodiment of the present invention, the spatial proximity criterion in step S4 adopts an adaptive Mahalanobis distance threshold judgment, the threshold of which is dynamically adjusted according to the target dynamic characteristics and the environmental noise level. In step S4, a partitioning processing strategy is adopted to divide the monitored airspace into multiple sub-regions, and the tracks in each sub-region are independently processed for aggregation.
[0015] In a preferred embodiment of the present invention, the state update after track aggregation in step S5 adopts a weighted fusion method, and the weights are determined according to the covariance matrix of each split track.
[0016] The motion consistency criteria in step S5 include:
[0017] The heading difference is less than the set threshold;
[0018] The speed difference is less than the set threshold;
[0019] The acceleration difference is less than the set threshold.
[0020] In a preferred embodiment of the present invention, the echo feature similarity criterion in step S5 includes:
[0021] The difference in radar cross section (RCS) is less than a set threshold;
[0022] The similarity of the echo amplitude distribution characteristics is greater than a set threshold;
[0023] The similarity of the echo fluctuation characteristics is greater than the set threshold.
[0024] In a preferred embodiment of the present invention, after completing step S6, it is also necessary to construct a memory and forgetting mechanism:
[0025] Preserve historical memory of temporarily split flight paths;
[0026] A forgetting mechanism is activated for split tracks that have not been re-associated within a set time.
[0027] The state vector of the Kalman filter in step S6 includes at least the target's position, velocity, and acceleration information.
[0028] In a preferred embodiment of the present invention, after completing step S5, a time persistence verification step is required, which requires the trajectory aggregation state to remain stable within N consecutive scanning cycles to confirm that the aggregation is effective.
[0029] In a preferred embodiment of the present invention, a processing system for a radar large target splitting track convergence algorithm based on a Kalman filter needs to be constructed before proceeding to step S1. The processing system includes a data acquisition module, a data preprocessing module, a Kalman filter bank, a correlation analysis module, and a track convergence module. The output of the data acquisition module is communicatively connected to the input of the data preprocessing module, the output of the data preprocessing module is communicatively connected to the input of the Kalman filter bank, the correlation analysis module and the track convergence module are bidirectionally connected, and the output of the Kalman filter bank is communicatively connected to the input of the correlation analysis module.
[0030] In a preferred embodiment of the present invention, the data acquisition module is used to acquire multi-frame monitoring data of the target by the radar, including the target's position information and velocity information;
[0031] The data preprocessing module filters the acquired radar data to remove noise interference. Common digital filtering algorithms, such as mean filtering or median filtering, can be used. The filtered data is then subjected to coordinate transformation to unify the data format and make it suitable for subsequent Kalman filter processing.
[0032] In a preferred embodiment of the present invention, the Kalman filter bank initializes a Kalman filter for each split track. The state vector of the Kalman filter includes state variables such as the target's position and velocity, and its measurement matrix is determined according to the radar's measurement model.
[0033] Kalman filtering is a recursive estimation method, meaning that as long as the estimated value of the state at the previous time step and the observed value of the current state are known, the estimated value of the current state can be calculated. Therefore, it does not need to record historical information of observation or estimation. The difference between Kalman filter and most filters is that it is a pure time-domain filter. Unlike frequency-domain filters such as low-pass filters, it does not need to be designed in the frequency domain and then converted to the time domain for implementation.
[0034] Kalman filtering is mainly divided into two parts: prediction and update. It requires the construction of the system's state equation and observation equation, and the initial state of the system is known. In the prediction phase, the filter uses the estimate of the previous state to make an estimate of the current state. In the update phase, the filter uses the observation value of the current state to optimize the prediction value obtained in the prediction phase to obtain a more accurate new estimate. Each Kalman filter predicts the trajectory according to its own state prediction equation and updates it according to the measurement update equation combined with new radar measurement data.
[0035] In a preferred embodiment of the present invention, the correlation analysis module calculates the correlation degree between each track. It can use a distance-based correlation algorithm, such as calculating the distance between the predicted positions of different tracks, and determining which split tracks may belong to the same large target based on the set correlation threshold. If the distance between two tracks is less than the threshold, they are considered to have a high correlation.
[0036] The track agglomeration module fuses the states of the Kalman filters of the associated split tracks. It can use a weighted average method to determine the weights based on the reliability of each track, and obtain a single agglomerated track. This track can more accurately reflect the true state of the large target.
[0037] The beneficial effects of this invention are as follows: This invention constructs a complete processing system by setting up a data acquisition module, a data preprocessing module, a Kalman filter bank, a correlation analysis module, and a track aggregation module. In actual use, the radar-detected points are initialized, an initial track file is established, and a Kalman filter is initialized for each track. The Kalman filter is used to predict the state of each track at the next moment. The newly obtained observation points are correlated and matched with the predicted state. The associated tracks that meet the spatial proximity criterion are aggregated and judged. The effectiveness of track aggregation is verified according to the motion consistency criterion and the echo feature similarity criterion. The aggregated observations that have passed the verification are used to update the state of the Kalman filter and output the aggregated stable track.
[0038] This invention improves the accuracy of radar convergence of split tracks for large targets through the above-mentioned combination, reducing the occurrence of misjudgments; enhances the stability of radar tracking of large targets, making subsequent track-based target analysis, command and decision-making more reliable; optimizes the use of radar resources, avoiding resource waste caused by erroneous multi-track processing; and manages, visualizes, and stores the above data and corresponding analysis results, which helps to realize radar large target split track convergence management through IoT cloud control, improving the intelligence level of radar large target split track convergence management. Attached Figure Description
[0039] For ease of explanation, the present invention will be described in detail below with reference to specific embodiments and accompanying drawings.
[0040] Figure 1 This is a flowchart illustrating the overall steps of a radar large target splitting track convergence method based on a Kalman filter according to the present invention.
[0041] Figure 2 This is a flowchart of a radar large target splitting trajectory convergence method based on Kalman filter according to the present invention;
[0042] Figure 3This is a simulation result of the Kalman filter-based trajectory convergence method for radar large target splitting based on the present invention.
[0043] Figure 4 This is an error diagram between the condensed track and the actual track of a radar large target splitting track condensation method based on Kalman filter according to the present invention. Detailed Implementation
[0044] like Figure 1 , Figure 2 , Figure 3 and Figure 4 As shown, the specific implementation adopts the following technical solution:
[0045] A radar large target split track convergence method based on Kalman filter includes the following specific steps:
[0046] S1. Initialize the points detected by the radar, create an initial track file, and initialize the Kalman filter for each track;
[0047] S2. Predict the state of each track at the next moment using a Kalman filter;
[0048] S3. Associate and match the newly obtained observation points with the predicted state;
[0049] S4. Perform agglomeration judgment on associated tracks that meet the spatial proximity criterion;
[0050] The spatial proximity criterion adopts an adaptive Mahalanobis distance threshold for judgment. The threshold is dynamically adjusted according to the target's dynamic characteristics and the level of environmental noise. In step S4, a partitioning processing strategy is adopted to divide the monitored airspace into multiple sub-regions and perform independent aggregation processing on the tracks in each sub-region.
[0051] S5. Verify the effectiveness of track convergence based on the motion consistency criterion and echo characteristic similarity criterion;
[0052] A weighted fusion method is used for the state update after track aggregation, and the weights are determined based on the covariance matrix of each split track.
[0053] The motion consistency criteria include: heading difference is less than a set threshold, velocity difference is less than a set threshold, and acceleration difference is less than a set threshold.
[0054] The echo feature similarity criteria include: radar cross section (RCS) difference is less than a set threshold, echo amplitude distribution feature similarity is greater than a set threshold, and echo fluctuation characteristic similarity is greater than a set threshold.
[0055] A time-duration verification step is also required, which requires the track aggregation state to remain stable over N consecutive scan cycles to confirm the aggregation is effective;
[0056] S6. Update the Kalman filter state using the validated condensed observations. The state vector of the Kalman filter should include at least the target's position, velocity, and acceleration information.
[0057] We also need to construct memory and forgetting mechanisms:
[0058] Preserve historical memory of temporarily split flight paths;
[0059] A forgetting mechanism is activated for split tracks that have not been re-associated within a set time.
[0060] S7, outputs the stable trajectory after condensation.
[0061] In a preferred embodiment of the present invention, a processing system for a radar large target splitting track convergence algorithm based on a Kalman filter needs to be constructed before proceeding to step S1. The processing system includes a data acquisition module, a data preprocessing module, a Kalman filter bank, a correlation analysis module, and a track convergence module. The output of the data acquisition module is communicatively connected to the input of the data preprocessing module, the output of the data preprocessing module is communicatively connected to the input of the Kalman filter bank, the correlation analysis module and the track convergence module are bidirectionally connected, and the output of the Kalman filter bank is communicatively connected to the input of the correlation analysis module.
[0062] The data acquisition module is used to collect multi-frame monitoring data of the radar on the target, including the target's location information (such as longitude, latitude, and altitude) and speed information;
[0063] The data preprocessing module filters the acquired radar data to remove noise interference. Common digital filtering algorithms, such as mean filtering or median filtering, can be used. The filtered data is then subjected to coordinate transformation to unify the data format and make it suitable for subsequent Kalman filter processing.
[0064] The Kalman filter bank initializes a Kalman filter for each split track. The state vector of the Kalman filter contains state variables such as the target's position and velocity, and its measurement matrix is determined according to the radar's measurement model.
[0065] Kalman filtering is a recursive estimation method, meaning that as long as the estimated value of the state at the previous time step and the observed value of the current state are known, the estimated value of the current state can be calculated. Therefore, it does not need to record historical information of observation or estimation. The difference between Kalman filter and most filters is that it is a pure time-domain filter. Unlike frequency-domain filters such as low-pass filters, it does not need to be designed in the frequency domain and then converted to the time domain for implementation.
[0066] Kalman filtering is mainly divided into two parts: prediction and update. It requires the construction of the system's state equation and observation equation, and the initial state of the system is known. In the prediction stage, the filter uses the estimate of the previous state to make an estimate of the current state. In the update stage, the filter uses the observation of the current state to optimize the prediction value obtained in the prediction stage to obtain a more accurate new estimate.
[0067] The state of the Kalman filter is represented by the following two variables:
[0068] 1) This represents the estimate of the state at time k;
[0069] 2) This represents the prediction of the state at time k, given the states at the past k-1 times.
[0070] 3) The posterior estimation error covariance matrix measures the accuracy of the estimated values.
[0071] The prediction-correction process is as follows:
[0072]
[0073] a) Prediction
[0074] In the prediction step, the state at the current moment is predicted based on the state and control variables at the previous moment. This predicted value is an estimate because it does not yet take into account the observed values at the current moment. The error covariance matrix of the predicted value is calculated by the error covariance matrix at the previous moment and the system noise covariance matrix.
[0075]
[0076] b) Update
[0077] In the update step, the state estimate at the current moment is calculated based on the observed and predicted values at the current moment. This estimate is a more accurate estimate because it has taken into account the observed values at the current moment. The error covariance matrix of the state estimate is calculated by using the error covariance matrix, observation noise covariance matrix and Kalman gain obtained in the prediction step.
[0078]
[0079] The above five formulas are the core formulas of Kalman filtering. The update steps are presented in a simpler and easier-to-understand form. First, calculate the following three quantities:
[0080]
[0081] Then use them to update the filter variables:
[0082]
[0083] The above formula is only valid for calculating the optimal Kalman gain.
[0084] Each Kalman filter predicts the trajectory based on its own state prediction equation and updates it based on the measurement update equation combined with new radar measurement data.
[0085] The correlation analysis module calculates the correlation between each track. It can use a distance-based correlation algorithm, such as calculating the distance between the predicted positions of different tracks. Based on the set correlation threshold, it determines which split tracks may belong to the same large target. If the distance between two tracks is less than the threshold, they are considered to have a high correlation.
[0086] The track agglomeration module fuses the states of the Kalman filters of the associated split tracks. It can use a weighted average method to determine the weights based on the reliability of each track (e.g., the covariance matrix elements of the Kalman filter) to obtain a single agglomerated track. This track can more accurately reflect the true state of the large target.
[0087] In practice:
[0088] 1. Data Acquisition: The radar scans the target area according to the set scanning cycle and transmits the target position and velocity information obtained from each scan to the data acquisition module;
[0089] 2. Preprocessing: The data acquisition module sends the data to the preprocessing module. The preprocessing module first performs mean filtering, for example, by averaging five consecutive frames of data to remove noise. Then, it performs coordinate transformation to convert the radar local coordinates to coordinates in the geodetic coordinate system.
[0090] 3. Kalman Filtering: The preprocessed data is input into the Kalman filters corresponding to each split track. The state prediction equation of each Kalman filter calculates the state prediction value for the next moment based on the target's motion model (such as uniform linear motion model or uniformly accelerated linear motion model). The measurement update equation then corrects the prediction value based on the new radar measurements.
[0091] 4. Correlation Analysis and Track Convergence: The correlation analysis module periodically (e.g., once every 10 frames of data) calculates the correlation degree between each track. When the correlation degree of two tracks meets the conditions, a fusion operation is performed in the track convergence module. For example, if track A and track B are correlated, the weights are calculated based on the covariance matrix of their Kalman filters during convergence. Then, the state vectors of the two tracks are weighted and averaged to obtain the state vector of the new converged track.
[0092] Simulation results explanation:
[0093] 1. Simulation Scene Setup
[0094] Target motion model: Uniform linear motion (initial position [0, 0], velocity [2m / s, 1m / s])
[0095] Radar observation characteristics:
[0096] The large target is split into 3 points. The main point is located near the actual location, and the split points are randomly offset within a range of ±1.8m.
[0097] Superimposed Gaussian observation noise (standard deviation 0.5m);
[0098] Kalman filter parameters:
[0099] State vector: [px, py, vx, vy]
[0100] The initial covariance P = diag([1, 1, 0.5, 0.5])
[0101] Process noise Q = diag([0.1, 0.1, 0.05, 0.05])
[0102] The observation noise R = diag([0.5, 0.5])
[0103] 2. Visualization of simulation results
[0104] (1) Combined with appendix Figure 3 As can be seen, the black solid line in the track comparison diagram represents the actual track (ideal motion path);
[0105] Red scatter dots: Split points observed by radar (3 points per moment);
[0106] Blue dashed line: The condensed trajectory (Kalman filter output);
[0107] Key observations:
[0108] The convergent track closely tracks the real track, effectively suppressing the random dispersion of split points;
[0109] During turns or acceleration changes (if present), the filter's prediction error may increase briefly but converge quickly.
[0110] (2) Position error analysis X / Y direction error curves:
[0111] Error range: ±0.48m (significantly lower than the original observation noise of 0.5m)
[0112] The error peak occurs when the target's motion direction changes abruptly (in this example, the motion is uniform, so the error is stable). Quantitative indicators:
[0113] RMSE (Root Mean Square Error):
[0114] X direction: 0.18m
[0115] Y direction: 0.22m
[0116] Maximum instantaneous error: 0.49m (during a period of observation with high noise).
[0117] Performance metrics:
[0118] Velocity estimation convergence time: 2 seconds (stabilizes after approximately 4 observation periods) Steady-state velocity error:
[0119] <0.05m / s
[0120] The results analysis and summary are shown in the table below:
[0121]
[0122] Analysis of track cohesion effect:
[0123] (1) Processing of split dots
[0124] Before aggregation: 3 points per second, with an average spacing of 2.2m (maximum 4.1m).
[0125] After aggregation: all split points are merged into a single track, and the association logic can be further optimized using Mahalanobis distance test;
[0126] (2) Algorithm robustness test
[0127] Test scenario: A fake dot (offset 3m) is manually added at t=10s, far from the main target.
[0128] result:
[0129] False points were successfully eliminated (due to failure to meet the motion consistency criterion); condensed tracks remained undisturbed; Key conclusions:
[0130] 1. Effectiveness: The algorithm achieves a coagulation success rate of over 95% when the spacing between splitting points is less than 10m and the speed is consistent.
[0131] 2. Real-time performance: Single filtering time < 0.1ms (MATLAB environment), meeting real-time processing requirements; 3. Limitations and improvement directions:
[0132] Limitations: If the spacing between split points is too large (>10m), it may be misjudged as multiple targets.
[0133] Improvement directions: Introduce an adaptive agglomeration threshold (dynamically adjusted according to the target size) and add multiple hypothesis tracking (MHT) to handle complex splitting scenarios.
[0134] MATLAB simulation code
[0135]
[0136]
[0137]
[0138]
[0139]
[0140] Specifically, in practical applications, multiple data acquisition modules are used in conjunction with a data preprocessing module, a Kalman filter bank, a correlation analysis module, and a track aggregation module. These multiple data acquisition modules are located in different geographical locations. This invention constructs a complete processing system by setting up data acquisition modules, data preprocessing modules, Kalman filter banks, correlation analysis modules, and track aggregation modules. In actual use, the radar-detected points are initialized, an initial track file is established, and a Kalman filter is initialized for each track. The Kalman filter is used to predict the state of each track at the next moment. The newly obtained observation points are correlated and matched with the predicted state. The associated tracks that meet the spatial proximity criterion are aggregated. The effectiveness of track aggregation is verified according to the motion consistency criterion and the echo feature similarity criterion. The Kalman filter state is updated using the verified aggregated observations, and the aggregated stable track is output.
[0141] This invention improves the accuracy of radar convergence of split tracks for large targets through the above-mentioned combination, reducing the occurrence of misjudgments; enhances the stability of radar tracking of large targets, making subsequent track-based target analysis, command and decision-making more reliable; optimizes the use of radar resources, avoiding resource waste caused by erroneous multi-track processing; and manages, visualizes, and stores the above data and corresponding analysis results, which helps to realize radar large target split track convergence management through IoT cloud control, improving the intelligence level of radar large target split track convergence management.
[0142] Those skilled in the art will recognize that the modules and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0143] The data acquisition, data preprocessing, Kalman filter bank, correlation analysis, and trajectory aggregation modules may or may not be physically separate. The components displayed as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0144] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program instructions, such as USB flash drives, portable hard drives, read-only storage servers, random access storage servers, magnetic disks, or optical disks.
[0145] Furthermore, it should be noted that the combination of the various technical features in this case is not limited to the combination methods described in the claims of this case or the combination methods described in the specific embodiments. All technical features described in this case can be freely combined or combined in any way, unless they contradict each other.
[0146] It should be noted that the above examples are merely specific embodiments of the present invention, and the present invention is obviously not limited to the above embodiments, with many similar variations. All modifications that can be directly derived or conceived by those skilled in the art from the content disclosed in this invention should fall within the protection scope of this invention.
[0147] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., 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 radar large target splitting track convergence method based on Kalman filter, characterized in that, The specific steps include the following: S1. Initialize the points detected by the radar, create an initial track file, and initialize the Kalman filter for each track; S2. Predict the state of each track at the next moment using a Kalman filter; S3. Associate and match the newly obtained observation points with the predicted state; S4. Perform agglomeration judgment on associated tracks that meet the spatial proximity criterion; S5. Verify the effectiveness of track convergence based on the motion consistency criterion and echo characteristic similarity criterion; S6. Update the Kalman filter state using the validated condensation observations; S7, outputs the stable trajectory after condensation.
2. The radar large target splitting track convergence method based on Kalman filter according to claim 1, characterized in that: The spatial proximity criterion in step S4 adopts an adaptive Mahalanobis distance threshold judgment, and the threshold is dynamically adjusted according to the target dynamic characteristics and environmental noise level. In step S4, a partitioning processing strategy is adopted to divide the monitored airspace into multiple sub-regions, and the tracks in each sub-region are independently processed for aggregation.
3. The radar large target splitting track convergence method based on Kalman filter according to claim 1, characterized in that: In step S5, a weighted fusion method is used for the state update after track aggregation, and the weights are determined based on the covariance matrix of each split track. The motion consistency criteria in step S5 include: The heading difference is less than the set threshold; The speed difference is less than the set threshold; The acceleration difference is less than the set threshold.
4. The radar large target splitting track convergence method based on Kalman filter according to claim 3, characterized in that: The echo feature similarity criteria in step S5 include: The difference in radar cross section (RCS) is less than a set threshold; The similarity of the echo amplitude distribution characteristics is greater than a set threshold; The similarity of the echo fluctuation characteristics is greater than the set threshold.
5. The radar large target splitting trajectory convergence method based on Kalman filter according to claim 1, characterized in that: After completing step S6, it is also necessary to construct a memory and forgetting mechanism: Preserve historical memory of temporarily split flight paths; A forgetting mechanism is activated for split tracks that have not been re-associated within a set time. The state vector of the Kalman filter in step S6 includes at least the target's position, velocity, and acceleration information.
6. The radar large target splitting track convergence method based on Kalman filter according to claim 5, characterized in that: After completing step S5, a time persistence verification step is required, which requires the track aggregation state to remain stable within N consecutive scan cycles to confirm that the aggregation is effective.
7. The radar large target splitting track convergence method based on Kalman filter according to claim 1, characterized in that: Before proceeding to step S1, a processing system for a radar large target splitting track convergence algorithm using a Kalman filter needs to be constructed. The processing system includes a data acquisition module, a data preprocessing module, a Kalman filter bank, a correlation analysis module, and a track convergence module. The output of the data acquisition module is communicatively connected to the input of the data preprocessing module, and the output of the data preprocessing module is communicatively connected to the input of the Kalman filter bank. The correlation analysis module and the track convergence module are bidirectionally connected, and the output of the Kalman filter bank is communicatively connected to the input of the correlation analysis module.
8. The radar large target splitting track convergence method based on Kalman filter according to claim 7, characterized in that: The data acquisition module is used to acquire multi-frame monitoring data of the target from the radar, including the target's position information and velocity information; The data preprocessing module filters the acquired radar data to remove noise interference.
9. The radar large target splitting track convergence method based on Kalman filter according to claim 8, characterized in that: The Kalman filter bank initializes a Kalman filter for each split track. The state vector of the Kalman filter contains state variables such as the target's position and velocity, and its measurement matrix is determined according to the radar's measurement model.
10. The radar large target splitting track convergence method based on Kalman filter according to claim 9, characterized in that: The correlation analysis module calculates the correlation degree between each track and determines which split tracks may belong to the same large target based on the set correlation threshold. The track fusion module fuses the states of the Kalman filters of correlated split tracks.