Multi-sensor hybrid information fusion method based on Bessel fitting feedback

Abstract

The present invention discloses a multi-sensor hybrid information fusion method based on Bessel fitting feedback, which belongs to the field of radar target data processing and is suitable for track association and target tracking in multi-sensor data fusion. The main implementation process is: using Bessel curve fitting in real time to perform secondary processing on the tracks of each sensor, using the similarity of the Bessel curve fitting results to perform track association judgment, and after the track association is successful, using Bessel curve fitting feedback to eliminate abnormal associated points and then perform hybrid fusion filtering on the target, which not only improves the success rate of track association but also improves the target tracking accuracy.

Description

Technical Field

[0001] The invention belongs to the field of radar target data processing. Background Art

[0002] With the increasing diversification and complexity of modern warfare and the continuous development of information fusion theory, the demand for target detection and tracking is also increasing. The ability of traditional single sensors to obtain battlefield situation and intelligence has certain limitations, and the means of single sensors in electronic confrontation are relatively simple. Through the networking of multiple sensors, the comprehensive detection of the entire combat airspace can be achieved, and the resources and information advantages of different sensors can be comprehensively utilized. The local tracks of the same target obtained by sensors of different systems, different frequency bands, and different polarization modes can be combined into a system fusion track, which can make the tracking accuracy of the fused track higher on different platforms, improve the stability of the entire combat system, and reduce the loss of tracking accuracy due to weather, electromagnetic interference or damage to a single sensor.

[0003] Therefore, it is necessary to effectively process a large amount of various sensor track information. Processing these sensor track information means increasing the amount of information to be processed, which is likely to involve contradictions and inconsistencies between the track data of each sensor. By adopting a suitable multi-sensor track fusion method, the performance differences and complementarity of each sensor can be used as much as possible to make up for the defects of a single sensor, and the track correlation can be comprehensively analyzed, so that the entire fusion system is fault-tolerant and the speed of track processing and the correctness of decision-making can be improved. Summary of the invention

[0004] The present invention provides a multi-sensor hybrid information fusion method based on Bessel fitting feedback, which can make full use of the track situation information of each sensor, improve the correctness and robustness of track association through the Bessel curve fitting method, and greatly improve the filtering accuracy of the fused track during hybrid fusion filtering.

[0005] The multi-sensor hybrid information fusion method based on Bessel fitting feedback proposed in the present invention includes:

[0006] Step 1: Perform Bezier curve fitting on multi-sensor track multi-cycle data respectively, and obtain the motion status of each track according to the curve fitting results;

[0007] Step 2: For the track matching the motion situation, calculate and retain the relevant parameters of the Bezier fitting curve, use the multi-scale curve parameters to perform similar fuzzy judgment, and judge whether the track is related based on the comprehensive similarity of the curve;

[0008] Step 3: After the track association is successful, a fused track is formed. The fused track is subjected to secondary Bezier curve fitting, and the fitting result is fed back to the fused track update link. The associated point tracks of the fused track are screened through the curve fitting feedback result.

[0009] Step 4: Use the filtered point tracks to update the fused track, and use the fitting curve as a benchmark to calculate the track accuracy, track filtering quality, and track threat level in real time.

[0010] Furthermore, the similarity fuzzy judgment of the Bezier fitting curve related parameters and the multi-scale curve parameters in step 2 includes:

[0011] ① The parameters related to the Bessel fitting curve include the heading C of the track curve at any time node T , Speed ​​V T 、Azimuth velocity V a , elevation angular velocity V e , Mobility Factor C v , direction ρ A , distance ρ R , elevation angle ρ E ;

[0012] ②Similar fuzzy judgment function is ∈=ω1C T +ω2V T +ω3V a +ω4V e +ω5C v +ω6ρ A +ω7ρ R +ω8ρ E , where ω1~ω N Adaptive adjustment based on curve fitting results.

[0013] Furthermore, the step 3 of screening the associated points of the fused track by the curve fitting feedback result includes: obtaining the relevant decision parameters of the filtering time node according to the fitting curve, and matching and screening the points in the wave gate according to the associated decision function δ:

[0014] δ=μ1ΔC T +μ2ΔV T +μ3ΔV a +μ4ΔV e +μ5ΔC v +μ6Δρ A +μ7Δρ R +μ8Δρ E

[0015] Among them, μ1~μ8 are weighting factors of relevant parameters.

[0016] The present invention makes full use of the track situation information of each sensor and combines the differences and complementarities of the tracks between sensors, which not only improves the correctness and stability of track association, but also improves the filtering accuracy of the fused track. At the same time, it can also utilize each other's non-interference information in complex electromagnetic environments, weather environments, and clutter environments to achieve stable detection of targets by the combat system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of multi-sensor hybrid fusion filtering based on Bessel curve fitting. DETAILED DESCRIPTION

[0018] After spatially transforming the tracks of multiple sensors, the present invention uses the Bezier curve fitting method to achieve the extension of the target track in time, fully considers the error jitter of the target moving under different sensors, combines multi-scale feature information for fuzzy association judgment, improves the accuracy of track association through fuzzy judgment of multiple membership degrees, and simultaneously performs secondary Bezier curve fitting on the fused track. The fitting result is used to screen the multi-sensor point tracks, eliminate abnormal associated point tracks, and improve the tracking accuracy of the fused track.

[0019] The steps of the embodiment of the present invention are as follows:

[0020] (1) Receive multi-sensor track information in real time, cache multi-source tracks, and send them to the track association module and fusion track filtering module for processing;

[0021] (2) Perform spatial registration on the multi-sensor tracks and map them to the location of the current processing center for spatial coordinate conversion, so that the multi-sensor tracks can be associated and judged in the same spatial coordinate system;

[0022] (3) Bezier curve fitting is performed on the multi-period data of multi-sensor tracks respectively, and the motion status of each track is obtained according to the curve fitting results. At the same time, the curve fitting results are used to mark the abnormal points of the original track received by the single sensor, and the non-abnormal points are used to perform secondary processing on the single-source original track to obtain more accurate track information;

[0023] (4) Before the track association, the track to be judged is time-aligned, and the target motion situation information obtained by separate curve fitting is used to predict the track to be compared to a unified time node;

[0024] (5) Multi-source tracks are judged by track association, and fuzzy association judgment is performed using the independent fitting parameters of single-source tracks. The specific parameters include the heading C at any time node. T , Speed ​​V T 、Azimuth velocity V a , elevation angular velocity V e , Mobility Factor Cv , direction ρ A , distance ρ R , elevation angle ρ E At the same time, in order to avoid the inaccuracy of a certain dimension information affecting the association result, the fuzzy association judgment method is used for association judgment. The fuzzy association function is ∈=ω1C T +ω2V T +ω3V a +ω4V e +ω5C v +ω6ρ A +ω7ρ P +ω8ρ E , where ω1~ω N Adaptive adjustment based on curve fitting results;

[0025] If the judgment is successful (∈≥γ, γ is the judgment threshold), the tracks are determined to be related;

[0026] (6) After the track association is successful, the multi-period point track sequence of the successfully associated multi-sensor track is used to perform quadratic Bezier curve fitting on the fused track, and the result of the quadratic curve fitting is used to provide feedback to the fused track. The stability of the fused track filtering point track is ensured by screening the point track;

[0027] According to the associated decision function δ, the points in the wave gate are matched and screened:

[0028] δ=μ1ΔC T +μ2ΔV T +μ3ΔV a +μ4ΔV e +μ5ΔC v +μ6Δρ A +μ7Δρ R +μ8Δρ E , where μ1~μ8 are

[0029] Weighting factors for relevant parameters;

[0030] The multi-period point sequence with an error accuracy lower than 3 times of the fitting result is subjected to subsequent filtering:

[0031] R j ={i|δ<3σ}

[0032] (7) The fused track is updated using the screened valid points, and the track accuracy, track filtering quality, and track threat level are calculated in real time based on the fitting curve.

Claims

1. A multi-sensor hybrid information fusion method based on Bessel fitting feedback, characterized by: Step 1: Perform Bezier curve fitting on multi-sensor track multi-cycle data respectively, and obtain the motion status of each track according to the curve fitting results; Step 2: Calculate and retain the relevant parameters of the Bessel fitting curve for the track matching of the motion situation, use the multi-scale curve parameters to perform similar fuzzy judgment, and judge whether the track is related based on the comprehensive similarity of the curve, including: ① The parameters related to the Bessel fitting curve include the heading C of the track curve at any time node T , Speed ​​V T 、Azimuth velocity V a , elevation angular velocity V e , Mobility Factor C v , direction ρ A , distance ρ R , elevation angle ρ E ; ②Similar fuzzy judgment function is ∈=ω1C T +ω2V T +ω3V a +ω4V e +ω5C v +ω6ρ A +ω7ρ R +ω8ρ E , where ω1~ω N Adaptive adjustment based on curve fitting results; Step 3: After the track association is successful, a fused track is formed. The fused track is subjected to quadratic Bezier curve fitting, and the fitting result is fed back to the fused track update link. The associated points of the fused track are screened through the curve fitting feedback result, including: obtaining the relevant decision parameters of the filtering time node according to the fitting curve, and matching and screening the points in the wave gate according to the associated decision function δ: δ=μ1ΔC T +μ2ΔV T +μ3ΔV a +μ4ΔV e +μ5ΔC v +μ6Dr A +μ7Dr R +μ8Dr E Among them, μ1~μ8 are weighting factors of relevant parameters; Step 4: Use the filtered points to update the fused track, and use the fitting curve as a benchmark to calculate the track accuracy, track filtering quality, and track threat level in real time.

Images