A track correction method based on the true probability of track-related points

Abstract

The present invention relates to radar data processing, and in particular to a track correction method based on the true probability of track-associated point tracks, which includes obtaining a radar point track of a current frame, performing point track condensation and point track-track association on the radar point track of the current frame, and generating a radar track of the current frame; performing track filtering on the radar track in the current frame radar track, and calculating a predicted value of the radar track in the next frame radar track; calculating the probability that the radar track-associated radar point track in the next frame radar track is a true target return point according to a measured value and a predicted value of the radar track in the next frame radar track, and calculating radar track data in the next frame radar track; performing track filtering on the radar track in the next frame radar track, and calculating an ideal position of the radar track in the current frame radar track; the technical solution provided by the present invention can effectively overcome the defects of low radar track accuracy caused by failure to effectively correct the radar track, and severe lateral jitter of the radar track.

Description

Technical Field

[0001] The invention relates to radar data processing, and in particular to a track correction method based on the true probability of track-associated point tracks. Background Art

[0002] Accurately measuring the distance and angle of the target is the basis of millimeter-wave radar detection, and the filter tracking estimation algorithm is the core content of target tracking. Due to the performance limitations of the millimeter-wave radar measurement system, the returned measurement data often has the characteristics of high distance accuracy and low angle accuracy. The tracking algorithm is the prerequisite for achieving target tracking, and the reliability and estimation accuracy of the filtering algorithm are determined by the filtering model. However, in the actual work of millimeter-wave radar, in order not to increase the difficulty of modeling the filtering model, the difference between distance accuracy and angle accuracy is often less considered, and the Kalman, , Common filtering algorithms such as , result in the target motion state after filtering often showing severe lateral jitter, and there is a large gap between the filtered track and the actual target track.

[0003] In addition, in the actual operation of millimeter-wave radar, due to the susceptibility to noise interference, interference from strong reflective objects, multipath interference and other factors, the radar track after track association often contains a certain probability of false track. When the radar track is associated with a false track, the traditional processing method is mostly to continue track filtering based on the false track, and there is no adjustment or elimination operation on the false track. This processing method cannot effectively correct the false track, which seriously reduces the accuracy of the radar track. Summary of the invention

[0004] In view of the above-mentioned shortcomings of the prior art, the present invention provides a track correction method based on the true probability of track-associated point tracks, which can effectively overcome the defects of the prior art that the radar track cannot be effectively corrected, resulting in low radar track accuracy, and severe lateral jitter of the radar track.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0006] A track correction method based on the true probability of track-related point tracks comprises the following steps:

[0007] S1. Obtain the radar point trace of the current frame, and perform point trace condensation and point trace track association on the radar point trace of the current frame to generate the radar track of the current frame;

[0008] S2, performing track filtering on the radar track in the current frame radar track, and calculating the predicted value of the radar track in the next frame radar track;

[0009] S3, according to the measured value and predicted value of the radar track in the next frame of radar track, calculate the probability that the radar track associated radar point track in the next frame of radar track is the real target return point, and calculate the radar track data in the next frame of radar track;

[0010] S4, performing track filtering on the radar track in the next frame of radar track, and calculating the ideal position of the radar track in the current frame of radar track;

[0011] S5. Calculate the probability that the radar track associated with the radar point track in the previous frame of radar track is a true target return point, and correct the radar track data in the current frame of radar track in combination with the ideal position of the radar track in the current frame of radar track.

[0012] Preferably, obtaining the radar point trace of the current frame in S1, and performing point trace condensation and point trace track association on the radar point trace of the current frame to generate the radar track of the current frame includes:

[0013] S11. Get the k Frame radar trace P k ={ P k 1 , P k 2 ,…, P k kN}, No. k Frame radar trace P k Middle i The radar trace is ;

[0014] in, R k i Radar trace P k i The distance value of Radar trace P k i The angle value of V k i Radar trace P k i The speed value, RCS k i Radar trace P k i The RCS value, kN For the kThe number of radar traces in the frame radar trace;

[0015] S12, radar trace P k i Converted to Cartesian coordinate system, we get ;

[0016] in, x k i ,y k i Radar traces P k i In the Cartesian coordinate system X, Y Axis distance:

[0017] ;

[0018] ;

[0019] S13, for k Frame radar trace P k All radar traces converted to the Cartesian coordinate system are condensed using the DBSCAN algorithm, and the nearest neighbor algorithm is used to associate the traces to generate the k Frame radar track T k ={ T k 1 , T k 2 ,…, T k tkN}, No. k Frame radar track T k Middle i The radar track is T k i ={ X k i , Y k i , xV k i , yV k i , TP k i};

[0020] in, X k i , Y k i Radar tracks T k i exist X, Y The distance in the axial direction, xV k i , yV k i Radar tracks T k i exist X, Y The speed in the axis direction, TP k i Radar track T k i The associated historical point sequence, , Radar track T k i In the k Frame radar track T k The radar traces associated with Radar track T k i In the k - 1 frame radar track T k-1 The radar traces associated with Radar track T k i In the k - 2 frames of radar tracks T k-2 The radar traces associated with Radar track T k i In the k - 3 frames of radar tracks T k-3 The radar traces associated with Radar track T k i In the k - 4 frames of radar tracks T k-4 The radar traces associated with the .

[0021] Preferably, performing track filtering on the radar track in the current frame radar track and calculating the predicted value of the radar track in the next frame radar track in S2 includes:

[0022] S21, radar track using five-point linear fitting regression T k i Perform track filtering based on radar tracks T k i The associated historical point sequence ,exist X, Y The least square method is used for linear fitting in the axial direction to calculate the corresponding linear fitting regression value:

[0023] ;

[0024] ;

[0025] ;

[0026] ;

[0027] in, b x k , a x k , b y k , a y k All are radar tracks T k i The linear fitting regression value of , Radar tracks T k i The historical point sequence TP k i In the k -5+ j Frame radar track T k-5+j The radar traces associated with X , Y Distance in the axial direction;

[0028] S22, using the radar track calculated in S21 T k i The linear fit regression value of b xk , a x k , b y k , a y k , calculate the k +1 frame radar track T k+1 Medium radar track The predicted value of :

[0029] ;

[0030] ;

[0031] ;

[0032] ;

[0033] in, , Radar tracks exist X, Y The predicted distance in the axis direction, , Radar tracks exist X, Y Predicted value of the speed in the axis direction.

[0034] Preferably, in S3, according to the measured value and predicted value of the radar track in the next frame of radar track, the probability that the radar track associated radar point track in the next frame of radar track is the real target return point is calculated, and the radar track data in the next frame of radar track is calculated, including:

[0035] S31, real-time reception k +1 frame radar trace , for k +1 frame radar trace P k+1 All radar traces converted to the Cartesian coordinate system are condensed using the DBSCAN algorithm, and the nearest neighbor algorithm is used to associate the traces, so that the radar track T k i ={ X k i , Y k i , xV k i ,yV k i , TP k i} Associate new radar traces Later it became a radar track ;

[0036] in, Radar track T k i The latest historical point sequence associated with , Radar track T k i In the k +1 frame radar track T k+1 New radar traces associated with Radar track T k i In the k Frame radar track T k The radar traces associated with Radar track T k i In the k - 1 frame radar track T k-1 The radar traces associated with Radar track T k i In the k - 2 frames of radar tracks T k-2 The radar traces associated with Radar track T k i In the k - 3 frames of radar tracks T k-3 The radar traces associated with

[0037] S32, radar track T k i In the k +1 frame radar track T k+1 New radar traces associated with As radar tracks T k iThe measured value of the radar track calculated in S22 The predicted value of As radar tracks T k i The predicted value of , calculate the newly associated radar trace The probability of returning a point as the true target :

[0038] ;

[0039] in, , Radar tracks T k i The latest historical point sequence In the k +1 frame radar track T k+1 New radar traces associated with exist X , Y The distance in the axial direction, L x , L y The nearest neighbor algorithm is X , Y The correlation threshold in the axis direction;

[0040] S33, using the radar track calculated in S22 The predicted value of , and the newly associated radar traces calculated in S32 The probability of returning a point as the true target , calculate radar track Radar track data :

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] in, , Radar tracks exist X, Y The distance in the axial direction, , Radar tracks existX, Y The speed in the axis direction, Radar track T k i The latest historical point sequence associated with , which is the radar track A sequence of associated historical points.

[0046] Preferably, in S4, performing track filtering on the radar track in the next frame of radar track and calculating the ideal position of the radar track in the current frame of radar track includes:

[0047] S41, repeat S21, use five-point linear fitting regression to calibrate the radar track Perform track filtering based on radar tracks The associated historical point sequence ,exist X, Y The least square method is used for linear fitting in the axial direction to calculate the corresponding linear fitting regression value:

[0048] ;

[0049] ;

[0050] ;

[0051] ;

[0052] in, b x k+1 , a x k+1 , b y k+1 , a y k+1 All are radar tracks The linear fitting regression value of , Radar tracks The historical point sequence In the k -4+ j Frame radar track T k-4+j The radar traces associated with X , Y Distance in the axial direction;

[0053] S42, based on the radar track calculated in S41 The linear fitting regression value is used to calculate the radar track under the straight line condition. T k i In the k Frame radar track T k Ideal location in iT k i ={ ixT k i , iyT k i}:

[0054] ;

[0055] ;

[0056] in, ixT k i , yT k i They are respectively the radar tracks under straight line conditions T k i In the k Frame radar track T k Zhongzai X, Y The ideal distance in the axis direction.

[0057] Preferably, in S5, the probability that the radar track associated with the radar point track in the previous frame of radar track is the real target return point is calculated, and the radar track data in the current frame of radar track is corrected in combination with the ideal position of the radar track in the current frame of radar track, including:

[0058] S51, k - 1 frame radar track T k-1 Medium radar track In the k Frame radar track T k New radar traces associated with tP k i As radar tracks The measured value is obtained by using the radar track calculated in S42 T k i In the k Frame radar track T k Ideal location in iTk i , calculate the newly associated radar traces tP k i The probability of returning a point as the true target p k i :

[0059] ;

[0060] in, x k i ,y k i Radar tracks The latest historical point sequence TP k i In the k Frame radar track T k New radar traces associated with tP k i exist X , Y The distance in the axial direction, L x , L y The nearest neighbor algorithm is X , Y The correlation threshold in the axis direction;

[0061] S52, using the radar track calculated in S42 T k i In the k Frame radar track T k Ideal location in iT k i , and the newly associated radar traces calculated in S51 tP k i The probability of returning a point as the true target p k i , for radar tracks T k i Radar track data T k i ={ X k i , Y ki , xV k i , yV k i , TP k i} Make corrections:

[0062] ;

[0063] .

[0064] Compared with the prior art, the track correction method based on the true probability of track-associated point tracks provided by the present invention has the following beneficial effects:

[0065] 1) In view of the characteristics of millimeter-wave radar, which has high distance accuracy and low angle accuracy, five-point linear fitting regression is used for track filtering in the rectangular coordinate system, which greatly weakens the lateral jitter of the radar track;

[0066] 2) The probability of whether the radar point track associated with the radar track is the real target return point is calculated twice, which can effectively reduce the adverse impact of the wrong point track association on the radar track and improve the stability of target tracking;

[0067] 3) Based on the probability of whether the radar point track associated with the radar track is the real target return point, the filter value of the radar track is corrected twice, which effectively improves the accuracy of the radar track. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0069] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] A track correction method based on the true probability of track-related points, such as Figure 1 As shown, S1, obtaining the radar point trace of the current frame, and performing point trace condensation and point trace track association on the radar point trace of the current frame to generate the radar track of the current frame, specifically including:

[0072] S11. Get the k Frame radar trace P k ={ P k 1 , P k 2 ,…, P k kN}, No. k Frame radar trace P k Middle i The radar trace is ;

[0073] in, R k i Radar trace P k i The distance value of Radar trace P k i The angle value of V k i Radar trace P k i The speed value, RCS k i Radar trace P k i The RCS value, kN For the k The number of radar traces in the frame radar trace;

[0074] S12, radar trace P k i Converted to Cartesian coordinate system, we get ;

[0075] in, x k i ,y k i Radar traces Pk i In the Cartesian coordinate system X, Y Axis distance:

[0076] ;

[0077] ;

[0078] S13, for k Frame radar trace P k All radar traces converted to the Cartesian coordinate system are condensed using the DBSCAN algorithm, and the nearest neighbor algorithm is used to associate the traces to generate the k Frame radar track T k ={ T k 1 , T k 2 ,…, T k tkN}, No. k Frame radar track T k Middle i The radar track is T k i ={ X k i , Y k i , xV k i , yV k i , TP k i};

[0079] in, X k i , Y k i Radar tracks T k i exist X, Y The distance in the axial direction, xV k i , yV k iRadar tracks T k i exist X, Y The speed in the axis direction, TP k i Radar track T k i The associated historical point sequence, , Radar track T k i In the k Frame radar track T k The radar traces associated with Radar track T k i In the k - 1 frame radar track T k-1 The radar traces associated with Radar track T k i In the k - 2 frames of radar tracks T k-2 The radar traces associated with Radar track T k i In the k - 3 frames of radar tracks T k-3 The radar traces associated with Radar track T k i In the k - 4 frames of radar tracks T k-4 The radar traces associated with the .

[0080] S2, performing track filtering on the radar track in the current frame radar track, and calculating the predicted value of the radar track in the next frame radar track, specifically including:

[0081] S21, radar track using five-point linear fitting regression T k i Perform track filtering based on radar tracks T k i The associated historical point sequence ,exist X, YThe least square method is used for linear fitting in the axial direction to calculate the corresponding linear fitting regression value:

[0082] ;

[0083] ;

[0084] ;

[0085] ;

[0086] in, b x k , a x k , b y k , a y k All are radar tracks T k i The linear fitting regression value of , Radar tracks T k i The historical point sequence TP k i In the k -5+ j Frame radar track T k-5+j The radar traces associated with X , Y Distance in the axial direction;

[0087] S22, using the radar track calculated in S21 T k i The linear fit regression value of b x k , a x k , b y k , a y k , calculate the k +1 frame radar track T k+1 Medium radar track The predicted value of :

[0088] ;

[0089] ;

[0090] ;

[0091] ;

[0092] in, , Radar tracks exist X, Y The predicted distance in the axis direction, , Radar tracks exist X, Y Predicted value of the speed in the axis direction.

[0093] S3, according to the measured value and predicted value of the radar track in the next frame of radar track, calculate the probability that the radar track associated radar point track in the next frame of radar track is the real target return point, and calculate the radar track data in the next frame of radar track, specifically including:

[0094] S31, real-time reception k +1 frame radar trace , for k +1 frame radar trace P k+1 All radar traces converted to the Cartesian coordinate system are condensed using the DBSCAN algorithm, and the nearest neighbor algorithm is used to associate the traces, so that the radar track T k i ={ X k i , Y k i , xV k i , yV k i , TP k i} Associate new radar traces Later it became a radar track ;

[0095] in, Radar track T k i The latest historical point sequence associated with , Radar track T k i In the k +1 frame radar track T k+1 New radar traces associated with Radar track T k i In the k Frame radar track T k The radar traces associated with Radar track T k i In the k - 1 frame radar track T k-1 The radar traces associated with Radar track T k i In the k -2 frames of radar tracks T k-2 The radar traces associated with Radar track T k i In the k - 3 frames of radar tracks T k-3 The radar traces associated with

[0096] S32, radar track T k i In the k +1 frame radar track T k+1 New radar traces associated with As radar tracks T k i The measured value of the radar track calculated in S22 The predicted value of As radar tracks T k i The predicted value of , calculate the newly associated radar trace The probability of returning a point as the true target :

[0097] ;

[0098] in, , Radar tracks T k i The latest historical point sequence In the k +1 frame radar track T k+1 New radar traces associated with exist X , Y The distance in the axial direction, L x , L y The nearest neighbor algorithm is X , Y The correlation threshold in the axis direction;

[0099] S33, using the radar track calculated in S22 The predicted value of , and the newly associated radar traces calculated in S32 The probability of returning a point as the true target , calculate radar track Radar track data :

[0100] ;

[0101] ;

[0102] ;

[0103] ;

[0104] in, , Radar tracks exist X, Y The distance in the axial direction, , Radar tracks exist X, Y The speed in the axis direction, Radar track T k i The latest historical point sequence associated with , which is the radar track A sequence of associated historical points.

[0105] S4, performing track filtering on the radar track in the next frame of radar track, and calculating the ideal position of the radar track in the current frame of radar track, specifically including:

[0106] S41, repeat S21, use five-point linear fitting regression to calibrate the radar track Perform track filtering based on radar tracks The associated historical point sequence ,exist X, Y The least square method is used for linear fitting in the axial direction to calculate the corresponding linear fitting regression value:

[0107] ;

[0108] ;

[0109] ;

[0110] ;

[0111] in, b x k+1 , a x k+1 , b y k+1 , a y k+1 All are radar tracks The linear fitting regression value of , Radar tracks The historical point sequence In the k -4+ j Frame radar track T k-4+j The radar traces associated with X , Y Distance in the axial direction;

[0112] S42, based on the radar track calculated in S41 The linear fitting regression value is used to calculate the radar track under the straight line condition. T k i In the k Frame radar track T k Ideal location in iT k i ={ ixT k i , iyT k i}:

[0113] ;

[0114] ;

[0115] in, ixT k i , yT k i They are respectively the radar tracks under straight line conditions T k i In the k Frame radar track T k Zhongzai X, Y The ideal distance in the axis direction.

[0116] S5, calculating the probability that the radar track associated with the radar point track in the previous frame of radar track is the real target return point, and correcting the radar track data in the current frame of radar track in combination with the ideal position of the radar track in the current frame of radar track, specifically including:

[0117] S51, k - 1 frame radar track T k-1 Medium radar track In the k Frame radar track T k New radar traces associated with tP k i As radar tracks The measured value is obtained by using the radar track calculated in S42 T k i In the k Frame radar track T k Ideal location in iT k i , calculate the newly associated radar traces tP k i The probability of returning a point as the true target p k i :

[0118] ;

[0119] in, x k i ,y k i Radar tracks The latest historical point sequence TP k i In the k Frame radar track T k New radar traces associated with tP k i exist X , Y The distance in the axial direction, L x , L y The nearest neighbor algorithm is X , Y The correlation threshold in the axis direction;

[0120] S52, using the radar track calculated in S42 T k i In the k Frame radar track T k Ideal location in iT k i , and the newly associated radar traces calculated in S51 tP k i The probability of returning a point as the true target p k i , for radar tracks T k i Radar track data T k i ={ X k i , Y k i , xV k i , yV k i , TP k i} Make corrections:

[0121] ;

[0122] .

[0123] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A track correction method based on the true probability of track association points, characterized by: The following steps are involved: S1. Obtain the radar point trace of the current frame, and perform point trace condensation and point trace track association on the radar point trace of the current frame to generate the radar track of the current frame; S2, performing track filtering on the radar track in the current frame radar track, and calculating the predicted value of the radar track in the next frame radar track; S3, according to the measured value and predicted value of the radar track in the next frame of radar track, calculate the probability that the radar track associated radar point track in the next frame of radar track is the real target return point, and calculate the radar track data in the next frame of radar track; S4, performing track filtering on the radar track in the next frame of radar track, and calculating the ideal position of the radar track in the current frame of radar track; S5, calculating the probability that the radar track associated with the radar point track in the previous frame of radar track is the real target return point, and correcting the radar track data in the current frame of radar track in combination with the ideal position of the radar track in the current frame of radar track; In S1, the radar point trace of the current frame is obtained, and the point trace condensation and point trace track association of the radar point trace of the current frame are performed to generate the radar track of the current frame, including: S11. Get the k Frame radar trace P k ={ P k 1 , P k 2 ,…, P k kN }, No. k Frame radar trace P k Middle i The radar trace is ; in, R k i Radar trace P k i The distance value of Radar trace P k i The angle value of V k i Radar trace P k i The speed value, RCS k i Radar trace P k i The RCS value, kN For the k The number of radar traces in the frame radar trace; S12, radar trace P k i Converted to Cartesian coordinate system, we get ; in, x k i ,y k i Radar traces P k i In the Cartesian coordinate system X, Y Axis distance: ； ； S13, for k Frame radar trace P k All radar traces converted to the Cartesian coordinate system are condensed using the DBSCAN algorithm, and the nearest neighbor algorithm is used to associate the traces to generate the k Frame radar track T k ={ T k 1 , T k 2 ,…, T k tkN }, No. k Frame radar track T k Middle i The radar track is T k i ={ X k i , Y k i , xV k i , Y k i , TP k i }; in, X k i , Y k i Radar tracks T k i exist X, Y The distance in the axial direction, xV k i , Y k i Radar tracks T k i exist X、 Y The speed in the axis direction, TP k i Radar track T k i The associated historical point sequence, , Radar track T k i In the k Frame radar track T k The radar traces associated with Radar track T k i In the k - 1 frame radar track T k-1 The radar traces associated with Radar track T k i In the k -2 frames of radar tracks T k-2 The radar traces associated with Radar track T k i In the k - 3 frames of radar tracks T k-3 The radar traces associated with Radar track T k i In the k - 4 frames of radar tracks T k-4 The radar traces associated with the .

2. The track correction method based on the true probability of track association points according to claim 1 is characterized by: In S2, the radar track in the current frame radar track is filtered, and the predicted value of the radar track in the next frame radar track is calculated, including: S21, radar track using five-point linear fitting regression T k i Perform track filtering based on radar tracks T k i The associated historical point sequence ,exist X, Y The least square method is used for linear fitting in the axial direction to calculate the corresponding linear fitting regression value: ； ； ； ； in, b x k , a x k , b y k , a y k All are radar tracks T k i The linear fitting regression value of , Radar tracks T k i The historical point sequence TP k i In the k -5+ j Frame radar track T k-5+j The radar traces associated with X , Y Distance in the axial direction; S22, using the radar track calculated in S21 T k i The linear fit regression value of b x k , a x k , b y k , a y k , calculate the k +1 frame radar track T k+1 Medium radar track The predicted value of : ； ； ； ； in, , Radar tracks exist X, Y The predicted distance in the axis direction, , Radar tracks exist X, Y Predicted value of the speed in the axis direction.

3. The track correction method based on the true probability of track association points according to claim 2 is characterized in that: In S3, according to the measured value and predicted value of the radar track in the next frame of radar track, the probability that the radar track associated radar point track in the next frame of radar track is the real target return point is calculated, and the radar track data in the next frame of radar track is calculated, including: S31, real-time reception k +1 frame radar trace , for k +1 frame radar trace P k+1 All radar traces converted to the Cartesian coordinate system are condensed using the DBSCAN algorithm, and the nearest neighbor algorithm is used to associate the traces, so that the radar track T k i ={ X k i , Y k i , xV k i , Y k i , TP k i } Associate new radar traces Later it became a radar track ; in, Radar track T k i The latest historical point sequence associated with , Radar track T k i In the k +1 frame radar track T k+1 New radar traces associated with Radar track T k i In the k Frame radar track T k The radar traces associated with Radar track T k i In the k - 1 frame radar track T k-1 The radar traces associated with Radar track T k i In the k -2 frames of radar tracks T k-2 The radar traces associated with Radar track T k i In the k - 3 frames of radar tracks T k-3 The radar traces associated with S32, radar track T k i In the k +1 frame radar track T k+1 New radar traces associated with As radar tracks T k i The measured value of the radar track calculated in S22 The predicted value of As radar tracks T k i The predicted value of , calculate the newly associated radar trace The probability of returning a point as the true target : ； in, , Radar tracks T k i The latest historical point sequence In the k +1 frame radar track T k+1 New radar traces associated with exist X , Y The distance in the axial direction, L x , L y The nearest neighbor algorithm is X , Y The correlation threshold in the axis direction; S33, using the radar track calculated in S22 The predicted value of , and the newly associated radar traces calculated in S32 The probability of returning a point as the true target , calculate radar track Radar track data : ； ； ； ； in, , Radar tracks exist X, Y The distance in the axial direction, , Radar tracks exist X, Y The speed in the axis direction, Radar track T k i The latest historical point sequence associated with , which is the radar track A sequence of associated historical points.

4. The track correction method based on the true probability of track-associated point tracks according to claim 3 is characterized by: In S4, the radar track in the next frame of radar track is filtered, and the ideal position of the radar track in the current frame of radar track is calculated, including: S41, repeat S21, use five-point linear fitting regression to calibrate the radar track Perform track filtering based on radar tracks The associated historical point sequence ,exist X, Y The least square method is used for linear fitting in the axial direction to calculate the corresponding linear fitting regression value: ； ； ； ； in, b x k+1 , a x k+1 , b y k+1 , a y k+1 All are radar tracks The linear fitting regression value of , Radar tracks The historical point sequence In the k -4+ j Frame radar track T k-4+j The radar traces associated with X , Y Distance in the axial direction; S42, based on the radar track calculated in S41 The linear fitting regression value is used to calculate the radar track under the straight line condition. T k i In the k Frame radar track T k Ideal location in iT k i ={ IX k i , iY k i }: ； ； in, IX k i ,yT k i They are respectively the radar tracks under straight line conditions T k i In the k Frame radar track T k Zhongzai X, Y The ideal distance in the axis direction.

5. The track correction method based on the true probability of track-associated point tracks according to claim 4 is characterized by: In S5, the probability that the radar track associated with the radar point track in the previous frame of radar track is the real target return point is calculated, and the radar track data in the current frame of radar track is corrected in combination with the ideal position of the radar track in the current frame of radar track, including: S51, k - 1 frame radar track T k-1 Medium radar track In the k Frame radar track T k New radar traces associated with Tp k i As radar tracks The measured value is obtained by using the radar track calculated in S42 T k i In the k Frame radar track T k Ideal location in iT k i , calculate the newly associated radar traces Tp k i The probability of returning a point as the true target p k i : ； in, x k i ,y k i Radar tracks The latest historical point sequence TP k i In the k Frame radar track T k New radar traces associated with Tp k i exist X , Y The distance in the axial direction, L x , L y The nearest neighbor algorithm is X , Y The correlation threshold in the axis direction; S52, using the radar track calculated in S42 T k i In the k Frame radar track T k Ideal location in iT k i , and the newly associated radar traces calculated in S51 Tp k i The probability of returning a point as the true target p k i , for radar tracks T k i Radar track data T k i ={ X k i , Y k i , xV k i , Y k i , TP k i } Make corrections: ； 。

Images