Target tracking method and device, equipment and storage medium

Abstract

The invention discloses a target tracking method and device, equipment and a storage medium. The method comprises the steps of S1, performing trajectory prediction according to a current frame image and a trajectory determined at a previous moment through a Kalman filter; s2, calculating a first track curvature of a historical track, a track at a previous moment and a predicted track at a current moment, an observation value of the historical track and a track obtained based on the current frame image, and a second track curvature of the predicted track at the current moment; s3, taking L2 norms of the first trajectory curvature and the second trajectory curvature as constraints to perform correlation calculation of the trajectory and the observation value, and matching the trajectory and the observation value according to a correlation calculation result; s4, updating the track which is successfully matched; and S5, taking the next frame of image as the current image, returning to the step S1 until the last frame of image, and outputting a final trajectory tracking result. The problem that in the prior art, target correlation is carried out based on Kalman filtering state estimation, the noise influence is large is solved.

Description

Technical Field

[0001] The present application relates to the field of target tracking technology, and in particular to a target tracking method, device, equipment and storage medium. Background Art

[0002] The matching strategy of traditional multi-target tracking technology usually uses Kalman filter to predict and analyze the target trajectory. Kalman filter is a linear estimator of discrete dynamical system in time domain. Kalman filter only needs the state estimate of the previous time step and the current observation to estimate the state of the next time step. Kalman filter maintains two variables: the posterior state estimate and the current observation. and the posterior estimated covariance matrix In target tracking tasks, the state transition model is generally used , observation model O, process noise , Kalman gain matrix K and observation noise To describe the filtering process. Assume that at time The observation value is , the prediction and update process of the traditional Kalman filter is as follows:

[0003]

[0004]

[0005] However, the traditional Kalman filter assumes that the target moves linearly, which is acceptable for short-term occlusion, but it is difficult to guarantee accuracy for long-term movement. The main reason is that the state predicted by the Kalman filter includes the speed in the x and y directions, and the variance of the speed is inversely proportional to the time interval. The smaller the interval, the greater the variance of the speed; and between consecutive frames of high-frame rate video, the noise of object displacement will cause drastic speed changes, thereby affecting the target tracking effect. Therefore, it is necessary to provide a noise reduction method to reduce the noise impact when performing target association based on Kalman filter state estimation. Summary of the invention

[0006] The present application provides a target tracking method, device, equipment and storage medium for improving the technical problem of the existing technology that target association based on Kalman filter state estimation is greatly affected by noise.

[0007] In view of this, the first aspect of the present application provides a target tracking method, comprising:

[0008] S1, predicting the trajectory at the current moment according to the current frame image and the trajectory determined at the previous moment through a Kalman filter to obtain the predicted trajectory at the current moment;

[0009] S2, calculating the first trajectory curvature of the historical trajectory of the preset time step, the trajectory of the previous moment and the predicted trajectory of the current moment, and the second trajectory curvature of the historical trajectory of the preset time step, the observation value of the trajectory obtained based on the current frame image and the predicted trajectory of the current moment;

[0010] S3, using the L2 norm of the first trajectory curvature and the second trajectory curvature as constraints to perform association calculation between the trajectory and the observation value, and matching the trajectory and the observation value according to the association calculation result;

[0011] S4, updating the successfully matched trajectory;

[0012] S5, taking the next frame image as the current image and returning to step S1 until the last frame image, outputting the final trajectory tracking result.

[0013] Optionally, the calculation formula of the trajectory curvature is:

[0014]

[0015] In the formula, is the trajectory curvature; t is the time, (x, y) is the trajectory information, is the time step of the historical trajectory.

[0016] Optionally, the using the L2 norm of the first trajectory curvature and the second trajectory curvature as a constraint to perform an association calculation between the trajectory and the observation value includes:

[0017] The distance between each trajectory and the observation value at the current moment is calculated, and a weighted sum is performed on the distance and the L2 norm of the first trajectory curvature and the second trajectory curvature to obtain an association calculation result of the trajectory and the observation value.

[0018] Optionally, updating the successfully matched trajectory includes:

[0019] Determine whether the successfully matched trajectory belongs to the lost trajectory that is retriggered;

[0020] If yes, the historical observations before the trajectory is lost and the associated observations after re-triggering are used as observation data, a high-order polynomial model is used to simulate the motion trajectory of the target, a simulated trajectory is obtained, and the matched trajectory is updated through a Kalman filter based on the simulated trajectory;

[0021] If not, the successfully matched trajectory is updated through the Kalman filter.

[0022] A second aspect of the present application provides a target tracking device, comprising:

[0023] A prediction unit, used to predict the trajectory at the current moment according to the current frame image and the trajectory determined at the previous moment through a Kalman filter, so as to obtain the predicted trajectory at the current moment;

[0024] A calculation unit, used to calculate a first trajectory curvature of a historical trajectory of a preset time step, a trajectory at a previous moment, and a predicted trajectory at a current moment, and a second trajectory curvature of a historical trajectory of a preset time step, an observation value of a trajectory obtained based on a current frame image, and a predicted trajectory at a current moment;

[0025] a matching unit, configured to perform an association calculation between a trajectory and an observation value by taking an L2 norm of the first trajectory curvature and the second trajectory curvature as a constraint, and matching the trajectory and the observation value according to the association calculation result;

[0026] An updating unit, used for updating the successfully matched trajectory;

[0027] The output unit is used to trigger the prediction unit by taking the next frame image as the current image until the last frame image, and output the final trajectory tracking result.

[0028] Optionally, the calculation formula of the trajectory curvature is:

[0029]

[0030] In the formula, is the trajectory curvature; t is the time, (x, y) is the trajectory information, is the time step of the historical trajectory.

[0031] Optionally, the calculation unit is specifically used to calculate the distance between each trajectory and the observation value at the current moment, and perform weighted summation on the distance and the L2 norm of the first trajectory curvature and the second trajectory curvature to obtain an association calculation result between the trajectory and the observation value.

[0032] Optionally, the updating unit is specifically used to:

[0033] Determine whether the successfully matched trajectory belongs to the lost trajectory that is retriggered;

[0034] If yes, the historical observations before the trajectory is lost and the associated observations after re-triggering are used as observation data to input into the high-order polynomial model to simulate the motion trajectory of the target, obtain the simulated trajectory, and update the successfully matched trajectory through the Kalman filter based on the simulated trajectory;

[0035] If not, the successfully matched trajectory is updated through the Kalman filter.

[0036] A third aspect of the present application provides an electronic device, the device comprising a processor and a memory;

[0037] The memory is used to store program code and transmit the program code to the processor;

[0038] The processor is used to execute the target tracking method described in any one of the first aspects according to the instructions in the program code.

[0039] A fourth aspect of the present application provides a computer-readable storage medium, which is used to store program code. When the program code is executed by a processor, it implements the target tracking method described in any one of the first aspects.

[0040] It can be seen from the above technical solutions that this application has the following advantages:

[0041] The target tracking method provided by the present application aims to improve the target tracking accuracy by introducing the trajectory curvature consistency calculated by the observation state and adding it to the matching matrix to reduce noise interference.

[0042] Furthermore, the present application proposes a simulated observation constraint method based on observation state, which back-simulates the observations during the target loss period based on the current observation values ​​and the observation values ​​before the loss, and re-updates the Kalman filter parameters based on the simulated observations to improve the cumulative error problem caused by the a posteriori update of the Kalman filter parameters based on state estimation under occlusion state, thereby further improving the target tracking accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 A flowchart of a target tracking method provided in an embodiment of the present application;

[0045] Figure 2 A schematic diagram of trajectory curvature provided in an embodiment of the present application;

[0046] Figure 3 Another flowchart of a target tracking method provided by an embodiment of the present application;

[0047] Figure 4 A schematic diagram of the structure of a target tracking device provided in an embodiment of the present application;

[0048] Figure 5 A structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0049] The present application provides a target tracking method, apparatus, device and storage medium for improving the problem of large noise influence in target association based on Kalman filter state estimation in the prior art.

[0050] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0051] For easier understanding, please refer to Figure 1 , the embodiment of the present application provides a target tracking method, comprising:

[0052] S1. Predict the trajectory at the current moment according to the current frame image and the trajectory determined at the previous moment through a Kalman filter to obtain the predicted trajectory at the current moment.

[0053] Perform target detection on the current frame image, obtain the detection frame of the target at the current time t, and thus obtain the observation value at the current time t . Through the Kalman filter according to the observation value at the current time t and the trajectory determined at the previous moment t-1 Predict the trajectory at the current time t and get the predicted trajectory at the current time t .

[0054] S2, calculating the first trajectory curvature of the historical trajectory of the preset time step, the trajectory of the previous moment and the predicted trajectory of the current moment, and the second trajectory curvature of the historical trajectory of the preset time step, the observation value of the trajectory obtained based on the current frame image and the predicted trajectory of the current moment.

[0055] Please refer to Figure 2 , this application adopts the historical trajectory , predict trajectory Respectively with the existing trajectory , current observation value The L2 norm of the trajectory curvature is used to represent the trajectory motion consistency. The trajectory curvature calculation formula is as follows:

[0056]

[0057] In the formula, is the trajectory curvature; the embodiment of the present application uses the center point coordinates (x, y) of the target to represent the trajectory information, is the time step of the historical trajectory.

[0058] The historical trajectory of the preset time step n , the trajectory determined at the previous moment t-1 and the predicted trajectory at the current time t Substitute into the above trajectory curvature calculation formula to calculate and obtain the first trajectory curvature ; Set the historical trajectory of the preset time step n , based on the observation value of the trajectory obtained from the current frame image and the predicted trajectory at the current time t Substitute into the above trajectory curvature calculation formula to calculate and obtain the second trajectory curvature .

[0059] S3. Using the L2 norm of the first trajectory curvature and the second trajectory curvature as constraints, perform association calculation between the trajectory and the observation value, and match the trajectory and the observation value according to the association calculation result.

[0060] In view of the problem of large noise influence in target association based on Kalman filter state estimation in traditional target tracking methods under high frame rate video, this application improves the traditional Kalman filter by introducing the trajectory curvature consistency of the observation state calculation and adding it to the matching matrix to reduce noise interference. Specifically, the distance between each trajectory and the observation value at the current moment is calculated, and the distance and the L2 norm of the first trajectory curvature and the second trajectory curvature are weighted and summed to obtain the association calculation result of the trajectory and the observation value. That is, when performing an association calculation (which can be an IOU association calculation) on the trajectory and the observation value at the current moment, the L2 norm of the first trajectory curvature and the second trajectory curvature is used as a constraint term. The association calculation formula is:

[0061]

[0062] In the formula, as the current object state estimation set; is the set of observations at the current new time step, is the weight factor; Contains observations of all existing trajectories; To calculate the distance between the observation set and the state estimation set at the current time step, is the L2 norm of the first trajectory curvature and the second trajectory curvature.

[0063] When the IOU association calculation method is adopted, matching is performed based on the IOU association calculation result. The specific matching process belongs to the existing technology and will not be described in detail here.

[0064] S4. Update the successfully matched trajectory.

[0065] In one embodiment, the existing Kalman filter may be directly used to update the successfully matched trajectory.

[0066] In another embodiment, considering the occlusion state, the rule of updating the Kalman filter parameters is to use the prior state estimation to perform a posteriori update, which is prone to error accumulation and thus affects the target tracking accuracy. To improve this problem, please refer to Figure 3 , the process of updating the successfully matched trajectory in this application includes:

[0067] S41, determining whether the successfully matched trajectory is a re-triggered lost trajectory;

[0068] S42: If yes, use the historical observation value of the re-triggered trajectory before it is lost. The observation value associated with the re-triggering is used as the observation data, a high-order polynomial model is used to simulate the motion trajectory of the target, a simulated trajectory is obtained, and a Kalman filter based on the simulated trajectory is used to update the trajectory that is successfully matched;

[0069] In order to solve the problem that the posterior update of Kalman filter parameters based on state estimation under occlusion will lead to cumulative errors, this application proposes a simulated observation constraint method based on observation state: by back-simulating the observations during the target loss period based on the current observation value and the observation value before the loss, and re-updating the Kalman filter parameters based on the simulated observations. Define the historical observations before the target is lost as , the target trajectory re-triggering association and the series of observations at the subsequent time are expressed as , ,..., , the embodiment of the present application introduces a third-order polynomial model to simulate the motion trajectory of the target:

[0070]

[0071] The prior art uses linear fitting to simulate the target's motion trajectory. This method does not take into account the complexity of the target's motion and has a poor simulation effect. This application expands the data range of the fitting to perform high-order polynomial fitting to describe more complex target motions and improve the simulation effect and accuracy. The third-order polynomial in the above formula can be of higher order, which can be selected adaptively according to the complexity of the motion and the amount of data.

[0072] According to the observed data, the least squares method is used to fit the curve and calculate the parameters. , , Then the simulated observations during the missing period can be defined as:

[0073]

[0074] The cycle of prediction and re-update is performed along the simulation trajectory, and the update formula is as follows:

[0075]

[0076] Where I is the unit matrix.

[0077] The lost track that is retriggered in the successfully matched track is updated by the above update formula, which reduces the influence of the error accumulated by the state estimation-based update in the occlusion state.

[0078] S43: If not, the successfully matched trajectory is updated through the Kalman filter.

[0079] For trajectories that are not re-triggered, trajectory updates are performed based on the traditional Kalman filter.

[0080] S5, taking the next frame image as the current image and returning to step S1 until the last frame image, outputting the final trajectory tracking result.

[0081] After all trajectories are updated at the current moment, the next frame of image is obtained for loop calculation until the last frame of image, and the latest trajectory update result is output. It should be noted that when pedestrian images are collected, the pedestrian images are from legally authorized sources or with the consent of the parties. When the collected pedestrian images are used for pedestrian tracking, they are mainly used for public safety.

[0082] This application addresses the problem of large noise influence in target association based on Kalman filter state estimation in traditional target tracking methods under high frame rate video. By introducing the trajectory curvature consistency calculated by the observation state and adding it to the matching matrix, noise interference is reduced and the target tracking accuracy is improved. Furthermore, this application proposes a simulated observation constraint method based on the observation state, which back-simulates the observations during the target loss period based on the current observation values ​​and the observation values ​​before the loss, and re-updates the Kalman filter parameters based on the simulated observations, so as to improve the cumulative error problem caused by the a posteriori update of the Kalman filter parameters based on the state estimation under occlusion state, and further improve the target tracking accuracy.

[0083] The above is an embodiment of a target tracking method provided by the present application, and the following is an embodiment of a target tracking device provided by the present application.

[0084] Please refer to Figure 4 , an object tracking device provided by an embodiment of the present application includes:

[0085] A prediction unit 410 is used to predict the trajectory at the current moment according to the current frame image and the trajectory determined at the previous moment through a Kalman filter to obtain the predicted trajectory at the current moment;

[0086] A calculation unit 420 is used to calculate a first trajectory curvature of a historical trajectory of a preset time step, a trajectory at a previous moment, and a predicted trajectory at a current moment, and a second trajectory curvature of a historical trajectory of a preset time step, an observation value of a trajectory obtained based on a current frame image, and a predicted trajectory at a current moment;

[0087] A matching unit 430, configured to use the L2 norm of the first trajectory curvature and the second trajectory curvature as constraints to perform association calculation between the trajectory and the observation value, and match the trajectory and the observation value according to the association calculation result;

[0088] An updating unit 440 is used to update the successfully matched trajectory;

[0089] The output unit 450 is used to trigger the prediction unit 410 by taking the next frame image as the current image until the last frame image, and output the final trajectory tracking result.

[0090] As a further improvement, the calculation formula of the trajectory curvature is:

[0091]

[0092] In the formula, is the trajectory curvature; t is the time, (x, y) is the trajectory information, is the time step of the historical trajectory.

[0093] As a further improvement, the calculation unit 420 is specifically used to calculate the distance between each trajectory and the observation value at the current moment, and perform weighted summation on the distance and the L2 norm of the first trajectory curvature and the second trajectory curvature to obtain the association calculation result of the trajectory and the observation value.

[0094] As a further improvement, the updating unit 440 is specifically configured to:

[0095] Determine whether the successfully matched trajectory belongs to the lost trajectory that is retriggered;

[0096] If yes, the historical observations before the trajectory is lost and the associated observations after re-triggering are used as observation data to input into the high-order polynomial model to simulate the motion trajectory of the target, obtain the simulated trajectory, and update the successfully matched trajectory through the Kalman filter based on the simulated trajectory;

[0097] If not, the successfully matched trajectory is updated through the Kalman filter.

[0098] The present application addresses the problem of large noise influence in target association based on Kalman filter state estimation in traditional target tracking methods under high frame rate video. The present application reduces noise interference and improves target tracking accuracy by introducing the trajectory curvature consistency calculated by observation state and adding it to the matching matrix. Furthermore, the present application performs back-tracking simulation of observations during the period of target loss based on current observation values ​​and observation values ​​before loss, and re-updates the Kalman filter parameters based on the simulated observations, so as to improve the cumulative error problem caused by the a posteriori update of the Kalman filter parameters based on state estimation under occlusion state, thereby further improving target tracking accuracy.

[0099] Please refer to Figure 5 , the embodiment of the present application further provides an electronic device, the device includes a processor 510 and a memory 520;

[0100] The memory 520 is used to store program codes and transmit the program codes to the processor 510;

[0101] The processor 510 is used to execute the target tracking method in the aforementioned method embodiment according to the instructions in the program code.

[0102] An embodiment of the present application also provides a computer-readable storage medium, which is used to store program code. When the program code is executed by a processor, the target tracking method in the aforementioned method embodiment is implemented.

[0103] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0104] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0105] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0106] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0107] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0108] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0109] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions for executing all or part of the steps of the method described in each embodiment of the present application through a computer device (which can be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store program codes.

[0110] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A target tracking method, characterized in that: include: S1, predicting the trajectory at the current moment according to the current frame image and the trajectory determined at the previous moment through a Kalman filter to obtain the predicted trajectory at the current moment; S2, calculating the first trajectory curvature of the historical trajectory of the preset time step, the trajectory of the previous moment and the predicted trajectory of the current moment, and the second trajectory curvature of the historical trajectory of the preset time step, the observation value of the trajectory obtained based on the current frame image and the predicted trajectory of the current moment; S3, using the L2 norm of the first trajectory curvature and the second trajectory curvature as constraints to perform association calculation between the trajectory and the observation value, and matching the trajectory and the observation value according to the association calculation result; S4, updating the successfully matched trajectory; S5, taking the next frame image as the current image and returning to step S1 until the last frame image, outputting the final trajectory tracking result.

2. The target tracking method according to claim 1, characterized in that: The calculation formula of the trajectory curvature is: In the formula, is the trajectory curvature; t is the time, (x, y) is the trajectory information, is the time step of the historical trajectory.

3. The target tracking method according to claim 1, characterized in that: The method of using the L2 norm of the first trajectory curvature and the second trajectory curvature as constraints to perform an association calculation between the trajectory and the observation value includes: The distance between each trajectory and the observation value at the current moment is calculated, and a weighted sum is performed on the distance and the L2 norm of the first trajectory curvature and the second trajectory curvature to obtain an association calculation result of the trajectory and the observation value.

4. The target tracking method according to claim 1, characterized in that: The updating of the successfully matched trajectory includes: Determine whether the successfully matched trajectory belongs to the lost trajectory that is retriggered; If yes, the historical observations before the trajectory is lost and the associated observations after re-triggering are used as observation data, a high-order polynomial model is used to simulate the motion trajectory of the target, a simulated trajectory is obtained, and the matched trajectory is updated through a Kalman filter based on the simulated trajectory; If not, the successfully matched trajectory is updated through the Kalman filter.

5. A target tracking device, characterized in that: include: A prediction unit, used to predict the trajectory at the current moment according to the current frame image and the trajectory determined at the previous moment through a Kalman filter, so as to obtain the predicted trajectory at the current moment; A calculation unit, used to calculate a first trajectory curvature of a historical trajectory of a preset time step, a trajectory at a previous moment, and a predicted trajectory at a current moment, and a second trajectory curvature of a historical trajectory of a preset time step, an observation value of a trajectory obtained based on a current frame image, and a predicted trajectory at a current moment; a matching unit, configured to perform an association calculation between a trajectory and an observation value by taking an L2 norm of the first trajectory curvature and the second trajectory curvature as a constraint, and matching the trajectory and the observation value according to the association calculation result; An updating unit, used for updating the successfully matched trajectory; The output unit is used to trigger the prediction unit by taking the next frame image as the current image until the last frame image, and output the final trajectory tracking result.

6. The target tracking device according to claim 5, characterized in that: The calculation formula of the trajectory curvature is: In the formula, is the trajectory curvature; t is the time, (x, y) is the trajectory information, is the time step of the historical trajectory.

7. The target tracking device according to claim 5, characterized in that: The calculation unit is specifically used to calculate the distance between each trajectory and the observation value at the current moment, and perform weighted summation on the distance and the L2 norm of the first trajectory curvature and the second trajectory curvature to obtain an association calculation result between the trajectory and the observation value.

8. The target tracking device according to claim 5, characterized in that: The updating unit is specifically used for: Determine whether the successfully matched trajectory belongs to the lost trajectory that is retriggered; If yes, the historical observations before the trajectory is lost and the associated observations after re-triggering are used as observation data to input into the high-order polynomial model to simulate the motion trajectory of the target, obtain the simulated trajectory, and update the successfully matched trajectory through the Kalman filter based on the simulated trajectory; If not, the successfully matched trajectory is updated through the Kalman filter.

9. An electronic device, characterized in that: The device comprises a processor and a memory; The memory is used to store program codes and transmit the program codes to the processor; The processor is used to execute the target tracking method according to any one of claims 1 to 5 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store program code, and when the program code is executed by a processor, the target tracking method according to any one of claims 1 to 5 is implemented.

Images