Multi-target tracking crossing method based on peak value matching

By using a combination method of peak matching and Kalman filters in multi-object tracking, the intersection state of the target is monitored and judged, and the interference and error problems during the intersection process are solved, and the accuracy and continuity of the intersection are improved.

CN119089212APending Publication Date: 2024-12-06THE 715TH RES INST OF CHINA SHIPBUILDING IND CORP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411038670.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The existing multi-objective tracking cross-over method is susceptible to interference from other nearby targets, resulting in errors in cross-state judgment, and errors in tracking may occur at the end of the cross.

Method used

Using a multi-objective tracking cross-section method based on peak matching, the target azimuth rate of change is obtained through smooth filtering of Kalman filter, the matching peak difference is monitored, the sustainability judgment of the same and different peaks is made, and the change rate is calculated in the time series for consistency. Comparison ensures the accuracy of the intersection state and the tracking continuity at the end.

Benefits of technology

It effectively reduces the interference of other targets on the intersection, improves the accuracy of the intersection process and tracking accuracy, and reduces the risk of follow-up errors at the end of the intersection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119089212A_ABST
    Figure CN119089212A_ABST
Patent Text Reader

Abstract

The invention discloses a multi-target tracking crossing method based on peak matching, and the method comprises the following steps: carrying out the smooth filtering of a target tracking azimuth through a Kalman filter in a non-crossing state, and obtaining the azimuth change rate of a target; after a cross command of any two targets i and j is issued, monitoring at a cross state moment by utilizing the azimuth difference of the matching peak values of the targets i and j; when entering a cross state, carrying out continuity judgment on the same peak by utilizing the matching peak values of the targets i and j; and continuously judging different peaks by using the matching peak values of the targets i and j, calculating the change rate # imgabs0 # of the azimuth of the peak value by using the time sequence of the azimuth of the matching peak value after the judgment passes, carrying out consistency comparison on the change rate # imgabs0 # and the azimuth change rate # imgabs1 # of the targets in a non-cross state, and ending the cross after the comparison passes. According to the method, the interference of other targets on the target in crossing can be reduced, and the tracking correctness is improved during crossing.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of multi-target tracking, and in particular relates to a multi-target tracking cross method based on peak matching. Background Art

[0002] When tracking multiple targets, it is often encountered that a strong energy target "takes away" a weak energy target, resulting in two or more targets (this article discusses two targets) affecting each other when their positions are close. Therefore, researchers have studied the problem of target crossing, and proposed a method of using the number of search peaks to judge the crossing state, and using the Kalman filter to process the crossing of the tracking target position during the crossing process. Specifically: 1. Issue a crossing command; 2. After the crossing command is issued, refer to the current position-energy result, take the current tracking position as the reference, and find the peak within a certain threshold. When the number of peaks is 2, the current two targets enter the crossing; 3. During the crossing, use the Kalman filter's position change rate to predict the target position; 4. Use the two target position values ​​as the reference, search for peaks, and record the crossing process in detail; 5. When the crossing process is coded as 2121 or the difference between the two target position values ​​meets a threshold, the crossing ends; 6. When the crossing ends, search for the peak near the current position value as the tracking position input after the crossing ends. There are two main problems that can be improved in this method: First, since the current target position is used as the reference to search for the number of nearby peaks within a certain range, but the target corresponding to each peak is not specifically pointed out, it may be interfered by other nearby targets except the two currently crossed targets. For example, when a cross command is issued to targets A and B, the current target A position is a, and the peak is searched within the θ degree range around a. At this time, there are targets other than the current two targets in this range. When searching for peaks, the number of peaks is not 2, and the cross is interfered and cannot enter the cross state; second, at the end of the cross, directly searching for nearby peaks for matching may result in matching errors, resulting in tracking errors after the cross.

[0003] Some researchers have also proposed a crossover method for peak matching, but the processing at the end of the crossover is somewhat rough, resulting in a high risk of following the wrong peak when crossing. Summary of the invention

[0004] The purpose of the present invention is to provide a multi-target tracking intersection method based on peak matching to solve the problem that the existing peak search intersection method may be interfered by other targets besides the two currently intersecting targets and the target is easy to be mistracked when the intersection ends.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A multi-target tracking intersection method based on peak matching, wherein the peak matching is to match the nearest peak value of the tracking target in a preset angle range in an azimuth-energy result based on the azimuth of the tracking target, and the multi-target intersection method comprises the following steps:

[0007] Step 1: In a non-crossing state, a Kalman filter is used to smooth the target tracking direction to obtain the target direction change rate;

[0008] Step 2, after issuing the cross command of any two targets i and j, the cross state is monitored by using the azimuth difference of the matching peaks of targets i and j, and step 3 is executed when entering the cross state;

[0009] Step 3, when entering the cross state, use the matching peak values ​​of targets i and j to determine the continuity of the same peak, and execute step 4 after the determination is passed;

[0010] Step 4: Use the matching peaks of targets i and j to determine the continuity of different peaks. After the determination is passed, use the time series of the matching peak position to calculate the change rate of the peak position. The rate of change of direction when the target is not intersecting Perform consistency comparison and end the crossover after the comparison passes.

[0011] Preferably, the peak value F of the match between two designated targets i and j is obtained during the monitoring of the cross state. i , F j Location Calculate the difference in the position of the matching peak When the difference is less than the preset threshold θ1 and the absolute value of the difference shows a decreasing trend, it is determined that the two designated targets enter a crossing state.

[0012] Preferably, the step of determining the decreasing trend comprises:

[0013] Calculate the difference sequence of the peak positions of the two targets i and j in the most recent time period When the rate of change of the difference sequence is negative, it is considered that the absolute value of the difference is decreasing.

[0014] Preferably, the continuity of the same peak is determined by: determining that the target i, j matches the peak value F i , F j Location exist And the duration is greater than t 2 When the judgment is passed.

[0015] Preferably, the persistence of the different peaks is determined as follows:

[0016] Get the matching peak F of two specified targets i and ji , F j Location exist And the duration reaches the time threshold t 21 When using the historical time t 21 The time series of the peak position is matched internally to calculate the rate of change of the peak position

[0017] Preferably, the consistency comparison is:

[0018] Determine the peak value F between two specified targets i and j i , F j Rate of change of position Whether it is in the corresponding target's azimuth change rate A certain error range ε pr If yes, the consistency comparison passes, otherwise it fails.

[0019] Preferably, before executing step 3, it is determined whether the tracking azimuth difference of the two targets after entering the crossing state is greater than a preset threshold, and if so, the crossing of the two targets is forced to end.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] The present invention can reduce the interference of other targets on the target in the intersection and improve the accuracy of tracking when leaving the intersection. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of the present invention.

[0023] Figure 2 is a flow chart of an embodiment of the present invention.

[0024] Figure 3 Schematic diagram of the azimuth-energy results and the tracking target.

[0025] Figure 4 This is a schematic diagram for judging whether to cross-enter or not.

[0026] Figure 5 Schematic diagram for judging the crossing into the "21" stage.

[0027] Figure 6 Schematic diagram for judging the crossing into the "212" stage. DETAILED DESCRIPTION

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 belong to the protection scope of the present invention.

[0029] The present invention provides a multi-target tracking and crossing method, which is based on the peak matching of the whole target crossing process, that is, continuously performing peak matching for each target from the non-crossing state to the end of crossing; the peak matching is based on the orientation of the tracking targets A, B, C, ..., N. In the azimuth-energy result, the nearest peak value is matched in the preset angle range for the tracking target A, B, C, ..., N. The matched nearest peak value is recorded as F A , F B , ..., F N ; The peak matching in different crossover stages has its corresponding preset angle range, and the preset angle range corresponding to the peak matching in different crossover stages is not necessarily the same.

[0030] Reference Figure 1 As shown, a multi-target tracking cross method based on peak matching includes the following four steps.

[0031] Step 1: In the non-intersecting state, start tracking the position of targets A, B, C, ..., N. In the azimuth-energy result, the target is tracked at the preset angle θ 0 Range matches the nearest peak F A , F B , ..., F N , while using the Kalman filter A , kalman B , kalman C , ..., kalman N Perform smooth filtering on the target tracking position to obtain the target position change rate

[0032] Figure 3 The figure shows the azimuth-energy result and the tracking target diagram at a certain moment. Figure 3 , at any time t in the non-crossing state 1 For example, the direction of the target A tracked at this moment is by Centered at the preset angles on the left and right Search for the nearest peak value within the A , at this time, it is considered that the target A and the peak F A Match, that is, t1 At this moment, target A corresponds to the peak value F in the azimuth-energy result. A ; at t 1 The position of target A will be tracked at all times As a Kalman filter A The input is passed through the Kalman filter A Get the target's azimuth change rate After a period of time, if the fluctuation error of the azimuth change rate is within a certain range, the azimuth change rate is considered to be stable, and the azimuth change rate at this time can be considered to be the azimuth change rate of the tracking target. Here, θ 0 The reference range is 1-4°.

[0033] Step 2, issue a cross command for any two targets i, j, i, j∈{A, B, ..., N}; after the cross command is issued, use the azimuth difference of the matching peaks of targets i, j to monitor the cross state at all times, and execute step 3 when entering the cross state.

[0034] Reference Figure 2 As shown, taking i=A and j=B as an example, combined with Figure 3 As shown, the step 2 includes the following sub-steps:

[0035] Step 2.1, issue a cross command for any two targets A and B;

[0036] Step 2.2: Get the matching peak value F of the two specified targets A and B in the current time window after the cross command is issued. A , F B Location Calculate the matching peak F A , F B The difference in the azimuth, that is, the peak angle difference Get the difference sequence of the two target matching peaks Determine whether the rate of change of the difference sequence is negative. If so, execute step 2.3, otherwise repeat step 2.2; here, the time window length is preset to t δ ;

[0037] Step 2.3: Determine the angle difference between the peak values ​​of the two targets A and B at the current moment Is it less than the angle threshold θ? 1 If yes, it is determined that the current moment has entered the cross state, and the cross coding process is recorded as "2", which is used to indicate that the two targets are each in one peak, with a total of two peaks. Execute step 3, otherwise return to step 2.2.

[0038] The present invention determines through step 2 whether the target orientation after entering the intersection changes linearly according to the orientation change rate before the intersection. If so, it is considered that the intersection state has been entered.

[0039] Step 3: When entering the cross state, use the matching peak values ​​of targets i and j to determine the continuity of the same peak, and execute step 4 after the determination is passed.

[0040] Combination Figure 5 As shown, after the two designated targets A and B enter the intersection state, it is considered that the change rate of the two target azimuths remains unchanged. The present invention makes a judgment through step 3, specifically: taking the tracking azimuths of the two targets A and B as the reference, searching for matching left and right preset angles θ 2 The nearest peak value F in the range A , F B , when the two matching peaks are located at If they are equal, the peaks where the two matching peaks are located are the same peak, and the same peak needs to be maintained for a preset time t 2 ; When the duration of the same peak reaches t 2 When , the cross-coding process is updated to "21", "1" represents that the peaks of the two targets are merged into one peak. Here, θ 2 The reference range is 2-4°, t 2 The reference range is 1-3s.

[0041] Step 4: Use the matching peaks of targets i and j to determine the continuity of different peaks. After the determination is passed, use the time series of the matching peak position to calculate the change rate of the peak position. The rate of change of direction when the target is not intersecting Perform consistency comparison and end after the comparison passes.

[0042] In step 4 of the present invention, targets i and j are at a preset angle θ 21 Search for the nearest peak value that matches the preset angle To preset the initial angle, is the azimuth change rate of the corresponding target, εpr is the error threshold of the azimuth change rate, t 交叉 The time to enter the crossover state.

[0043] After the two targets are fused into one peak, according to the normal crossover process, the next fused peak will be divided into two peaks as the crossover process proceeds. Figure 6 As shown, step 4 specifically includes the following steps:

[0044] Step 4.1, based on the tracking azimuths of the two targets, search for the nearest peak within the preset angle range, calculate whether the angle difference between the two peaks is not 0, record the azimuths of the respective matching peaks, and execute step 4.2, otherwise delete the azimuths of the matching peaks in the historical records and repeat step 4.1;

[0045] Step 4.2: Determine whether the recording duration meets the time threshold t 21 If yes, then use the time series of recorded azimuths to calculate the rate of change of the azimuth at the peak And execute step 4.3, otherwise execute step 4.1 Here, the time threshold t 21 The reference range is 2-4S.

[0046] Step 4.3 Determine the rate of change of the peak position Tracking target position change rate The preset azimuth change rate error threshold ε pr within, that is and If so, update the cross-coding process to "212", the two targets are matched with the current peak respectively, and the cross-state is released, otherwise return to step 4.1.

[0047] In this embodiment, ε pr Take 20%-40%.

[0048] Furthermore, before executing step 3, the tracking position difference of the two targets after entering the cross state is determined. Is it greater than the preset threshold θ d , where θ d Generally, the crossing angle threshold θ is taken 1 1.5 times of Force the crossover to end to prevent abnormal situations from causing the inability to exit the crossover state.

[0049] In the present invention, no matter whether the forced end of the crossover is after the end of the crossover in step 4, in order to ensure the continuity of target tracking after the end of the crossover, it is still necessary to match the peak value for each target.

[0050] In the present invention, it should be noted that during the implementation of the above steps, the positions of the tracking targets A, B, C, ..., N are continuously tracked. In the azimuth-energy result, the nearest peak is matched in the preset angle range for the tracked targets A, B, C, ..., N, rather than just for the two specified targets under the cross command. It is just that the azimuth of the specified target, the matching peak and its azimuth need to be extracted during the cross process to determine each cross stage.

[0051] The present invention matches the azimuth with the peak value during the peak search process. Since the peak value and the target are matched, the judgment of the crossing stage during the crossing process is more accurate and is not easily affected by the peak values ​​of other targets. In addition, since the tracking target azimuth is updated at a fixed azimuth change rate after entering the crossing, a large error may occur between the actual range and the tracking azimuth when exiting the crossing. Therefore, the matched peak value is observed for a period of time to try to ensure that the actual azimuth of the target is found after exiting the crossing, thereby increasing the probability of successful crossing.

[0052] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-target tracking cross method based on peak matching, characterized in that: The peak matching is to match the nearest peak value of the tracking target in a preset angle range in the azimuth-energy result based on the azimuth of the tracking target. The multi-target intersection method includes the following steps: Step 1: In a non-crossing state, a Kalman filter is used to smooth the target tracking direction to obtain the target direction change rate; Step 2: After issuing the cross command between any two targets i and j, the cross state is monitored by using the azimuth difference of the matching peaks of targets i and j. When entering the cross state, step 3 is executed; Step 3, when entering the cross state, use the matching peak values ​​of targets i and j to determine the continuity of the same peak, and execute step 4 after the determination is passed; Step 4: Use the matching peaks of targets i and j to determine the continuity of different peaks. After the determination is passed, use the time series of the matching peak position to calculate the change rate of the peak position. The rate of change of direction when the target is not intersecting Perform consistency comparison and end the crossover after the comparison passes.

2. A multi-target tracking cross method based on peak matching as claimed in claim 1, characterized in that: In the monitoring of the cross state, the peak value F of the two specified targets i and j is obtained. i , F j Location Calculate the difference in the position of the matching peak When the difference is less than the preset threshold θ1 and the absolute value of the difference shows a decreasing trend, it is determined that the two designated targets enter a crossing state.

3. A multi-target tracking cross method based on peak matching as claimed in claim 2, characterized in that: The step of determining the decreasing trend comprises: Calculate the difference sequence of the peak positions of the two targets i and j in the most recent time period When the rate of change of the difference sequence is negative, it is considered that the absolute value of the difference is decreasing.

4. A multi-target tracking cross method based on peak matching as claimed in claim 1, characterized in that: The persistence of the same peak is determined as follows: determine that the target i, j matches the peak value F i , F j Location exist And when the duration is greater than t2, it is judged as passed.

5. The multi-target tracking cross method based on peak matching as claimed in claim 1, characterized in that: The persistence of the different peaks is determined as follows: Get the matching peak F of two specified targets i and j i , F j Location exist And the duration reaches the time threshold t 21 When using the historical time t 21 The time series of the peak position is matched internally to calculate the rate of change of the peak position 6. A multi-target tracking cross method based on peak matching as claimed in claim 1, characterized in that: The consistency comparison is: Determine the peak value F between two specified targets i and j i , F j Rate of change of position Whether it is in the corresponding target's azimuth change rate A certain error range ε pr If yes, the consistency comparison passes, otherwise it fails.

7. The multi-target tracking cross method based on peak matching as claimed in claim 1, characterized in that: Before executing step 3, it is determined whether the tracking azimuth difference of the two targets after entering the crossing state is greater than a preset threshold. If so, the crossing of the two targets is forced to end.