Target interruption trajectory association method based on fusion strategy

Abstract

The invention relates to the technical field of computers. The target interruption trajectory association method based on the fusion strategy comprises the following steps: carrying out average distance calculation on a predicted position sequence and a position sequence of a new trajectory set to obtain an average distance value, and constructing a distance loss matrix based on the average distance value; performing feature parameter extraction processing on the historical trajectory and the new trajectory in the coarse correlation trajectory pair set to obtain a speed feature, an acceleration feature and an angular velocity feature; performing characteristic parameter loss calculation on the speed characteristic, the acceleration characteristic and the angular speed characteristic to generate a characteristic parameter loss matrix; performing weighted fusion processing on the distance loss matrix and the characteristic parameter loss matrix to generate an overall association cost matrix; and carrying out Hungary algorithm optimization processing on the overall association cost matrix, and outputting an association trajectory pair so as to achieve the technical effects of improving association robustness in a long-interruption scene, reducing the probability of misassociation of a high-maneuvering target and enhancing scene generalization ability.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method for associating target interruption trajectories based on a fusion strategy. Background Technology

[0002] With the rapid development of intelligent sensing technology, mobile target trajectory tracking is playing an increasingly important role in fields such as security monitoring, intelligent transportation, and unmanned system navigation. Trajectory association, as a core component of multi-target tracking, aims to reconnect fragmented trajectory segments due to occlusion, signal interruption, or other reasons into a continuous trajectory, thereby providing a data foundation for tasks such as behavior analysis and situation prediction.

[0003] Traditional techniques employ distance association methods based on trajectory prediction for trajectory association. However, this method suffers from accumulated prediction errors and insufficient robustness in scenarios with long interruptions. Parametric association methods utilizing motion features rely on the similarity of motion features before and after the interruption to achieve association, but they are prone to misassociations due to feature differences when facing highly maneuverable targets. Furthermore, traditional techniques often employ a single association strategy without considering the dynamic impact of interruption time on the association strategy, resulting in poor scenario generalization ability. Summary of the Invention

[0004] Therefore, it is necessary to provide a target interruption trajectory association method based on a fusion strategy to address the above-mentioned technical problems, so as to improve the association robustness in long interruption scenarios, reduce the probability of misassociation of highly maneuverable targets, and enhance the scenario generalization ability.

[0005] Firstly, this application provides a target interruption trajectory association method based on a fusion strategy, the method comprising:

[0006] Distance constraint filtering is performed on the historical trajectory set and the new trajectory set to obtain a set of coarsely correlated trajectory pairs;

[0007] The historical trajectories in the coarse correlation trajectory set are processed for trajectory prediction to generate a predicted location sequence;

[0008] The average distance between the predicted position sequence and the position sequence of the new trajectory set is calculated to obtain the average distance value, and a distance loss matrix is ​​constructed based on the average distance value; feature parameters are extracted from the historical trajectories and new trajectories in the coarse correlation trajectory set to obtain velocity features, acceleration features, and angular velocity features.

[0009] Feature parameter loss is calculated for velocity, acceleration, and angular velocity features to generate a feature parameter loss matrix; the distance loss matrix and the feature parameter loss matrix are weighted and fused to generate an overall association cost matrix; the overall association cost matrix is ​​optimized using the Hungarian algorithm to output associated trajectory pairs.

[0010] In one embodiment, a distance constraint filtering process is performed on the historical trajectory set and the new trajectory set to obtain a set of coarsely correlated trajectory pairs, including:

[0011] Obtain historical trajectory set and new trajectory set ,in , Representing historical trajectory The A location point, Indicates new trajectory The A location point, Representing historical trajectory The Location points coordinate, Representing historical trajectory The Location points coordinate, Representing historical trajectory The Location points coordinate;

[0012] Historical trajectory set The finish line With the new trajectory set starting position Perform Euclidean distance calculation to obtain the distance value. ,in, Representing historical trajectory The timestamp of the endpoint location, Representing historical trajectory End point coordinate, Representing historical trajectory End point coordinate, Indicates new trajectory starting point coordinate, Indicates new trajectory starting point coordinate, Indicates new trajectory The timestamp of the starting point;

[0013] Based on the preset target speed For distance value Perform a feasibility assessment of the motion and output that the motion is satisfied. The set of coarse correlation trajectories, where, Indicates the distance value. Indicates the target's maximum speed. Indicates the starting time of the new trajectory. It indicates the endpoint of the historical trajectory.

[0014] In one embodiment, the overall association cost matrix is ​​optimized using the Hungarian algorithm to output associated trajectory pairs, including:

[0015] Construct the objective function: ,in For the number of historical trajectories, For the number of new trajectories, These are the elements of the overall association cost matrix. Represent decision variables;

[0016] Set constraints:

[0017]

[0018] and ;

[0019] Solve for the assignment matrix that satisfies the constraints. ;

[0020] According to the allocation matrix China satisfies The elements output historical trajectory With new trajectory The associated trajectory pairs.

[0021] In one embodiment, feature parameter extraction processing is performed on historical and new trajectories in the coarse correlation trajectory pair set to obtain velocity features, acceleration features, and angular velocity features, including:

[0022] On the trajectory of history The end The historical trajectory is obtained by performing least-squares fitting on a sequence of location points. Directional velocity components and Directional velocity components ;

[0023] right and The velocity characteristics of the historical trajectories were obtained through synthetic processing. ;

[0024] The final historical trajectory Calculate the velocity of adjacent points from a sequence of location points to generate a velocity sequence of historical trajectories. ;

[0025] For velocity sequences Least squares fitting was performed to obtain the acceleration characteristics of the historical trajectory. ;

[0026] The final historical trajectory Calculate the orientation angles of adjacent points from a sequence of location points to generate an orientation angle sequence of the historical trajectory. ;

[0027] Opposite angle sequence Least squares fitting was performed to obtain the angular velocity characteristics of the historical trajectory. ;

[0028] New trajectory The front The sequence of position points is fitted using the least squares method to obtain the new trajectory. Directional velocity components and Directional velocity components ;

[0029] right and The velocity characteristics of the new trajectory are obtained through synthesis processing. ;

[0030] Foremost of the new trajectory Calculate the velocity sequence of adjacent points from a sequence of location points to generate a velocity sequence for a new trajectory. ;

[0031] For velocity sequences The acceleration characteristics of the new trajectory are obtained by performing least squares fitting. ;

[0032] Foremost of the new trajectory Calculate the orientation angles of adjacent points from a sequence of location points to generate a new trajectory orientation angle sequence. ;

[0033] Opposite angle sequence The angular velocity characteristics of the new trajectory are obtained by performing least squares fitting. ;

[0034] Velocity characteristics include the velocity characteristics of historical trajectories. and the velocity characteristics of the new trajectory Acceleration characteristics include acceleration characteristics of historical trajectories. and acceleration characteristics of the new trajectory Angular velocity characteristics include the angular velocity characteristics of historical trajectories. Angular velocity characteristics of the new trajectory .

[0035] In one embodiment, trajectory prediction processing is performed on historical trajectories in the coarse correlation trajectory pair set to generate a predicted location sequence, including:

[0036] Historical trajectories in the coarse correlation trajectory pair set The latest Points Perform a quadratic polynomial fitting and solve for the fitting coefficients. ;

[0037] Based on fitting coefficients For the new trajectory set time domain Perform location prediction and generate predicted location sequences. .

[0038] In one embodiment, the distance loss matrix and the feature parameter loss matrix are weighted and fused to generate an overall association cost matrix, including:

[0039] For each historical trajectory in the coarse correlation trajectory pair set Calculate the interruption time length based on the new trajectory j. ,in For the historical trajectory The final moment, This marks the starting point of the new trajectory.

[0040] Based on preset attenuation factor and interruption duration Preset attenuation factor Based on experimental data, the dynamic weighting coefficients were calculated. ;

[0041] Calculate complementary weighting coefficients ;

[0042] For distance loss matrix Middle elements With feature parameter loss matrix Middle elements Perform weighted fusion:

[0043]

[0044] Iterate through all coarsely correlated trajectory pairs and fuse the results of each trajectory pair. Fill in the matrix to generate the overall association cost matrix. .

[0045] In one embodiment, the historical trajectories in the coarse correlation trajectory pair set are... The latest Points Perform a quadratic polynomial fitting and solve for the fitting coefficients. ,include:

[0046] Obtain historical trajectories from the coarse correlation trajectory pair set The latest A sequence of positions of points, where the position information of each point is... , For the historical trajectory Total points;

[0047] For position sequence The relationship between coordinates and time is fitted using a quadratic polynomial to obtain... Directional fitting polynomial: ,in for Directional fitting coefficients;

[0048] For position sequence The relationship between coordinates and time is fitted using a quadratic polynomial to obtain... Directional fitting polynomial: ,in for Directional fitting coefficients;

[0049] based on Directional fitting coefficients and Directional fitting coefficients, obtaining the fitting coefficients .

[0050] Secondly, this application also provides a target interruption trajectory association system based on a fusion strategy, the system comprising:

[0051] The coarse correlation prediction module is used to perform distance constraint filtering on the historical trajectory set and the new trajectory set to obtain a set of coarse correlation trajectory pairs;

[0052] The historical trajectories in the coarse correlation trajectory set are processed for trajectory prediction to generate a predicted location sequence;

[0053] The loss feature construction module is used to calculate the average distance between the predicted position sequence and the position sequence of the new trajectory set, obtain the average distance value, and construct the distance loss matrix based on the average distance value; it also performs feature parameter extraction processing on the historical trajectories and new trajectories in the coarse correlation trajectory pair set to obtain velocity features, acceleration features, and angular velocity features.

[0054] The fusion decision module is used to calculate the feature parameter loss of velocity, acceleration and angular velocity features to generate the feature parameter loss matrix; to perform weighted fusion processing on the distance loss matrix and the feature parameter loss matrix to generate the overall association cost matrix; and to perform Hungarian algorithm optimization processing on the overall association cost matrix to output the associated trajectory pairs.

[0055] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods in the first aspect of this application.

[0056] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the methods in the first aspect of this application.

[0057] This application provides a target interruption trajectory association method based on a fusion strategy, which includes: performing distance constraint screening on the historical trajectory set and the new trajectory set to exclude trajectories that are not related to each other in order to simplify the calculation range; then generating a position sequence by predicting the coarsely related historical trajectories; and combining the position sequence of the new trajectory to calculate the average distance to construct a distance loss matrix, which effectively alleviates the error accumulation problem of simple trajectory prediction in long interruption scenarios and thus improves the robustness of association.

[0058] The system extracts velocity, acceleration, and angular velocity features from historical and new trajectories and calculates the loss to construct a feature parameter loss matrix. This captures the target's motion characteristics, and even with significant feature changes before and after interruptions in highly maneuverable targets, it can reduce the probability of misassociations through feature difference identification. The distance loss matrix and the feature parameter loss matrix are weighted and fused to generate an overall association cost matrix. Optimal matching is achieved using the Hungarian algorithm, integrating the advantages of both association strategies. This approach is suitable for distance matching requirements in short-interruption scenarios and can handle complex situations such as high maneuverability and long interruptions, thereby enhancing scenario generalization capabilities. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0060] Figure 1 This is a flowchart of a target interruption trajectory association method based on a fusion strategy, as described in one embodiment of the present invention.

[0061] Figure 2 This is a structural diagram of a target interruption trajectory association system based on a fusion strategy, according to one embodiment of the present invention. Detailed Implementation

[0062] To make the above-mentioned objects, features, and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0063] First, the application scenarios of the embodiments of this application are described. In the embodiments of this application, a target interruption trajectory association method based on a fusion strategy is provided, applicable to, but not limited to, security monitoring trajectory tracking, intelligent transportation vehicle trajectory completion, and continuous trajectory association for unmanned system navigation.

[0064] In illustrative purposes, the target interruption trajectory association method based on the fusion strategy provided in this application embodiment can also be applied to other application scenarios that require the re-association of interrupted trajectories, such as multi-sensor tracking of military targets, target trajectory connection in intelligent park inspection, and continuous identification of port logistics vehicle trajectories. This is only an example and does not limit the specific application scenarios.

[0065] like Figure 1 As shown, this application provides a target interruption trajectory association method based on a fusion strategy, the method comprising:

[0066] S101: Perform distance constraint filtering on the historical trajectory set and the new trajectory set to obtain a coarsely correlated trajectory pair set; perform trajectory prediction processing on the historical trajectories in the coarsely correlated trajectory pair set to generate a predicted position sequence.

[0067] For example, the location and time information of each historical trajectory in the historical trajectory set and the location and time information of each new trajectory in the new trajectory set are obtained. The endpoint position of each historical trajectory and the starting position of each new trajectory are determined. The spatial distance between them is calculated. The interval between the preset target maximum speed and the endpoint time of the historical trajectory and the starting time of the new trajectory is combined to determine whether the spatial distance is within a reasonable range of motion within the target time interval. The trajectory combinations that meet the constraints are retained to obtain a set of coarsely correlated trajectory pairs.

[0068] The latest valid position point sequence of each historical trajectory in the coarse correlation trajectory pair set is selected, and the sequence is fitted using a preset fitting algorithm to construct a prediction model. Based on the time domain range of the new trajectory set, the prediction position of the historical trajectory at each moment in the time domain is calculated by the prediction model, and all prediction positions are arranged in chronological order to generate a prediction position sequence.

[0069] S102: Calculate the average distance between the predicted position sequence and the position sequence of the new trajectory set to obtain the average distance value, and construct the distance loss matrix based on the average distance value; extract feature parameters from the historical trajectories and new trajectories in the coarse correlation trajectory set to obtain velocity features, acceleration features and angular velocity features.

[0070] For example, temporal registration is performed between the predicted position sequence and the position sequence of the new trajectory set to ensure a one-to-one correspondence in their time dimensions. The spatial distance between the predicted position and the new trajectory position at the corresponding time is calculated and averaged to obtain the average distance value. Based on the sequence correspondence between historical trajectories and new trajectories, the average distance values ​​of each coarsely correlated trajectory pair are filled into a matrix to construct a distance loss matrix. The latest valid position point sequence of the historical trajectory and the earlier valid position point sequence of the new trajectory are selected from the coarsely correlated trajectory pair set and processed using a preset fitting algorithm: Velocity features are obtained by solving for and synthesizing the velocity components from the position sequences; acceleration features are obtained by calculating and fitting the velocity sequences; and angular velocity features are obtained by calculating and fitting the direction angle sequences.

[0071] S103: Calculate the feature parameter loss for velocity, acceleration, and angular velocity features to generate a feature parameter loss matrix; perform weighted fusion processing on the distance loss matrix and the feature parameter loss matrix to generate an overall association cost matrix; optimize the overall association cost matrix using the Hungarian algorithm to output associated trajectory pairs.

[0072] For example, for each pair of historical and new trajectories in the coarsely correlated trajectory pair set, the degree of difference in their velocity, acceleration, and angular velocity characteristics is calculated. After normalizing these differences, the feature parameter loss values ​​for each trajectory pair are obtained by integrating them according to preset feature weights. The feature parameter loss values ​​are then filled into a matrix according to the sequence correspondence between historical and new trajectories to generate a feature parameter loss matrix. The interruption time length for each pair of historical and new trajectories is calculated. Dynamic weight coefficients and complementary weight coefficients are determined based on preset attenuation factors. The corresponding elements of the distance loss matrix are weighted using the dynamic weight coefficients, and the corresponding elements of the feature parameter loss matrix are weighted using the complementary weight coefficients. The sum of the two weighted element values ​​is then filled into the matrix according to their positions to generate the overall correlation cost matrix.

[0073] Construct a function that minimizes the total association cost, setting constraints that each historical trajectory can be associated with at most one new trajectory and each new trajectory can only be associated with one historical trajectory. Solve the assignment matrix that satisfies the constraints using the Hungarian algorithm, and output the associated trajectory pairs based on the historical trajectory index and the new trajectory index corresponding to the elements with a value of 1 in the assignment matrix.

[0074] An embodiment of this application provides a target interruption trajectory association method based on a fusion strategy, which includes: performing distance constraint screening on historical trajectory sets and new trajectory sets to exclude trajectories that may not be associated in order to simplify the calculation range; then generating a position sequence by predicting the coarsely associated historical trajectories; and combining the position sequence of the new trajectory to calculate the average distance and construct a distance loss matrix, which effectively alleviates the error accumulation problem of simple trajectory prediction in long interruption scenarios and thus improves the robustness of association.

[0075] The system extracts velocity, acceleration, and angular velocity features from historical and new trajectories and calculates the loss to construct a feature parameter loss matrix. This captures the target's motion characteristics, and even with significant feature changes before and after interruptions in highly maneuverable targets, it can reduce the probability of misassociations through feature difference identification. The distance loss matrix and the feature parameter loss matrix are weighted and fused to generate an overall association cost matrix. Optimal matching is achieved using the Hungarian algorithm, integrating the advantages of both association strategies. This approach is suitable for distance matching requirements in short-interruption scenarios and can handle complex situations such as high maneuverability and long interruptions, thereby enhancing scenario generalization capabilities.

[0076] In one embodiment, a distance constraint filtering process is performed on the historical trajectory set and the new trajectory set to obtain a set of coarsely correlated trajectory pairs, including:

[0077] (1) Obtain the historical trajectory set and new trajectory set ,in , Representing historical trajectory The A location point, Indicates new trajectory The A location point, Representing historical trajectory The Location points coordinate, Representing historical trajectory The Location points coordinate, Representing historical trajectory The Location points coordinate.

[0078] For example, explicitly set historical trajectory The composition of each historical trajectory From multiple location points composition, The value range is 1 to ( For the historical trajectory (Total number of points), each location point Include coordinate , coordinate Time coordinates At the same time, the new trajectory set was clearly defined. The composition of each new trajectory From multiple location points composition, The value range is 1 to ( For a new trajectory (Total number of points), each location point Include coordinate , coordinate Time coordinates The historical trajectory set can be obtained completely through the above methods. and new trajectory set All location point information.

[0079] Among them, historical trajectory collection It contains multiple historical trajectories, and the number of location points in each historical trajectory can vary; the new trajectory set It contains multiple new trajectories, and the number of location points on each new trajectory can be different.

[0080] (2) For historical trajectory sets The finish line With the new trajectory set starting position Perform Euclidean distance calculation to obtain the distance value. ,in, Representing historical trajectory The timestamp of the endpoint location, Representing historical trajectory End point coordinate, Representing historical trajectory End point coordinate, Indicates new trajectory starting point coordinate, Indicates new trajectory starting point coordinate, Indicates new trajectory The timestamp of the starting point.

[0081] For example, for historical trajectory sets Each historical trajectory Determine its endpoint location The endpoint is the historical trajectory. The Location points, including coordinate 1 , coordinate 1 Time coordinates For the new trajectory set Each new trajectory Determine its starting position The starting point is the new trajectory. The first location point, containing coordinate , coordinate Time coordinates After that, for each pair of historical trajectories... The finish line And new trajectory starting position Calculate the Euclidean distance between the two in the spatial dimension to obtain the corresponding distance value. .

[0082] The endpoint of the historical trajectory is the last position point of that trajectory; the starting point of the new trajectory is the first position point of that trajectory.

[0083] (3) Based on the preset target speed For distance value Perform a feasibility assessment of the motion and output that the motion is satisfied. The set of coarse correlation trajectories, where, Indicates the distance value. Indicates the target's maximum speed. Indicates the starting time of the new trajectory. It indicates the endpoint of the historical trajectory.

[0084] For example, for each historical trajectory And new trajectory Calculate the new trajectory from the pairs of trajectories formed. The starting moment With historical trajectory The end moment The time difference. Multiply this time difference by the preset target maximum speed. This yields the maximum distance the target can move within that time difference. Then, the distance value is... Compare with this maximum distance; if the distance value... If the distance is less than or equal to the maximum distance, the trajectory pair is deemed to satisfy the motion feasibility requirement and is included in the coarsely correlated trajectory pair set. After traversing all combinations of historical trajectories and new trajectories, the coarsely correlated trajectory pair set consisting of all trajectory pairs that satisfy the conditions is output.

[0085] Among them, the target maximum speed It is the maximum speed of motion preset based on the target's motion characteristics; the time difference is the difference between the starting time of the new trajectory and the ending time of the historical trajectory.

[0086] In one embodiment, the overall association cost matrix is ​​optimized using the Hungarian algorithm to output associated trajectory pairs, including:

[0087] (1) Construct the objective function: ,in For the number of historical trajectories, For the number of new trajectories, These are the elements of the overall association cost matrix. Represents decision variables.

[0088] For example, the explicit historical trajectory set includes A historical trajectory, the new trajectory set contains A new trajectory, each element in the overall correlation cost matrix Corresponding historical trajectory With new trajectory Association cost, decision variables Used to characterize historical trajectories With new trajectory Whether a connection is formed. All historical trajectories. From 1 to All new trajectories From 1 to corresponding and Multiply the products and sum them to construct an objective function that measures the total cost of association.

[0089] Among them, the number of historical trajectories It refers to the number of historical trajectories within a historical trajectory set; the number of new trajectories. It represents the number of new trajectories in the new trajectory set; elements of the overall association cost matrix. It is the trajectory of history And new trajectory Cost quantification during association; decision variables It can be 0 or 1, where 1 indicates that the two are related and 0 indicates that they are not related.

[0090] (2) Set constraints:

[0091]

[0092] and .

[0093] For example, for each new trajectory (from 1 to ), requiring all historical trajectories (from 1 to ) corresponding decision variables The sum equals 1, meaning each new trajectory can only be associated with one historical trajectory. For each historical trajectory... (from 1 to ), requiring all new trajectories (from 1 to ) corresponding decision variables The sum equals 1, meaning each historical trajectory can only be associated with one new trajectory. Meanwhile, the decision variables... The value range is limited to 0 or 1 to ensure the determinism of the association relationship.

[0094] The purpose of the constraints is to ensure a one-to-one matching relationship between historical trajectories and new trajectories, avoiding many-to-one or one-to-many situations.

[0095] (3) Solve for the assignment matrix that satisfies the constraints. .

[0096] For example, the Hungarian algorithm is applied to perform calculations based on the constructed objective function and set constraints. This algorithm obtains an allocation matrix with elements taking values ​​of 0 or 1 by performing a series of transformations and matching operations on the elements of the overall association cost matrix. The position of each element in the matrix corresponds to a historical trajectory. And new trajectory A value of 1 indicates that the two are associated, while a value of 0 indicates that they are not associated.

[0097] The Hungarian algorithm is a classic algorithm for solving task allocation problems, capable of finding the optimal allocation scheme that satisfies the constraints in polynomial time; the allocation matrix... It is OK A matrix of columns, the values ​​of which reflect the correlation between historical trajectories and new trajectories.

[0098] (4) Based on the allocation matrix China satisfies The elements output historical trajectory With new trajectory The associated trajectory pairs.

[0099] For example, traversing the allocation matrix one by one Each element, when a certain element is found When the value is 1, the historical trajectory corresponding to that element is determined. And new trajectory If a correlation exists, the historical trajectory and the new trajectory are combined into a correlated trajectory pair and output, until the entire allocation matrix has been traversed. Given all elements, obtain all associated trajectory pairs.

[0100] Among them, the associated trajectory pair is a combination of historical trajectories and new trajectories that are determined to have an association relationship after optimization by the Hungarian algorithm, and it is the final output of this step.

[0101] In one embodiment, feature parameter extraction processing is performed on historical and new trajectories in the coarse correlation trajectory pair set to obtain velocity features, acceleration features, and angular velocity features, including:

[0102] (1) Historical trajectory The end The historical trajectory is obtained by performing least-squares fitting on a sequence of location points. Directional velocity components and Directional velocity components .

[0103] right and The velocity characteristics of the historical trajectories were obtained through synthetic processing. .

[0104] The final historical trajectory Calculate the velocity of adjacent points from a sequence of location points to generate a velocity sequence of historical trajectories. .

[0105] For velocity sequences Least squares fitting was performed to obtain the acceleration characteristics of the historical trajectory. .

[0106] The final historical trajectory Calculate the orientation angles of adjacent points from a sequence of location points to generate an orientation angle sequence of the historical trajectory. .

[0107] Opposite angle sequence Least squares fitting was performed to obtain the angular velocity characteristics of the historical trajectory. .

[0108] For example, regarding historical trajectories Select its last A sequence of location points, containing multiple location points at the end of the historical trajectory. coordinate, Coordinate and time information. Least squares fitting is performed on the sequence of location points to establish location (…). , By understanding the mathematical relationship between time and the historical trajectory, and minimizing the sum of squared errors, we can obtain the historical trajectory. Directional velocity components and Directional velocity components. Vector synthesis is performed on the two velocity components to obtain the velocity characteristics of the historical trajectory. The final historical trajectory... Given a sequence of position points, the velocity between adjacent points is calculated (derived from the displacement and time difference of adjacent points), generating a velocity sequence for the historical trajectory. This velocity sequence is then subjected to least-squares fitting to establish the relationship between velocity and time, yielding the acceleration characteristics of the historical trajectory. Finally, the historical trajectory is... Given a sequence of position points, calculate the orientation angle between adjacent position points (the angle between the line connecting adjacent points and the reference direction) to generate a sequence of orientation angles for the historical trajectory. Perform least-squares fitting on this sequence of orientation angles to establish the relationship between orientation angles and time, thereby obtaining the angular velocity characteristics of the historical trajectory.

[0109] Among them, least squares fitting is used to determine the fitting parameters of each feature; the position point sequence is the set of positions and time series of the ends of the historical trajectory; the velocity sequence is the set of time series of the motion velocities of adjacent points; and the orientation angle sequence is the set of time series of the orientation angles of adjacent points.

[0110] (2) For the new trajectory The front The sequence of position points is fitted using the least squares method to obtain the new trajectory. Directional velocity components and Directional velocity components .

[0111] right and The velocity characteristics of the new trajectory are obtained through synthesis processing. .

[0112] Foremost of the new trajectory Calculate the velocity sequence of adjacent points from a sequence of location points to generate a velocity sequence for a new trajectory. ;

[0113] For velocity sequences The acceleration characteristics of the new trajectory are obtained by performing least squares fitting. ;

[0114] Foremost of the new trajectory Calculate the orientation angles of adjacent points from a sequence of location points to generate a new trajectory orientation angle sequence. ;

[0115] Opposite angle sequence The angular velocity characteristics of the new trajectory are obtained by performing least squares fitting. ;

[0116] Velocity characteristics include the velocity characteristics of historical trajectories. and the velocity characteristics of the new trajectory Acceleration characteristics include acceleration characteristics of historical trajectories. and acceleration characteristics of the new trajectory Angular velocity characteristics include the angular velocity characteristics of historical trajectories. Angular velocity characteristics of the new trajectory .

[0117] For example, for the new trajectory Select its first one A sequence of location points, containing multiple location points of the initial segment of the new trajectory. coordinate, Coordinate and time information. Least squares fitting is performed on the sequence of location points to establish location (…). , By understanding the mathematical relationship between time and the new trajectory, and minimizing the sum of squared errors, we can obtain the new trajectory. Directional velocity components and Directional velocity components. Vector synthesis is performed on the two velocity components to obtain the velocity characteristics of the new trajectory. The foremost part of the new trajectory... Given a sequence of position points, calculate the velocity between adjacent points (derived from the displacement and time difference of adjacent points) to generate a velocity sequence for the new trajectory. Perform least-squares fitting on this velocity sequence to establish the relationship between velocity and time, obtaining the acceleration characteristics of the new trajectory. Then, for the initial position of the new trajectory... Given a sequence of position points, calculate the orientation angle between adjacent position points (the angle between the line connecting adjacent points and the reference direction) to generate a new trajectory orientation angle sequence. Perform least-squares fitting on this orientation angle sequence to establish the relationship between orientation angle and time, thus obtaining the angular velocity characteristics of the new trajectory.

[0118] Among them, least squares fitting is used to determine the fitting parameters of each feature; the position point sequence is the set of positions and time series of the starting segment of the new trajectory; the velocity sequence is the set of time series of the motion velocities of adjacent points; and the orientation angle sequence is the set of time series of the orientation angles of adjacent points.

[0119] In one embodiment, trajectory prediction processing is performed on historical trajectories in the coarse correlation trajectory pair set to generate a predicted location sequence, including:

[0120] (1) Historical trajectories in the coarse correlation trajectory pair set The latest Points Perform a quadratic polynomial fitting and solve for the fitting coefficients. .

[0121] For example, for historical trajectories in the set of coarse correlation trajectory pairs Select its latest These points contain the ends of historical trajectories. each position coordinate, Coordinates and time coordinates. (Specifically for...) The relationship between coordinates and time To understand the relationship between coordinates and time, a quadratic polynomial model is constructed. This model is then solved by minimizing the sum of squared errors between the actual coordinates and the model's predicted coordinates for each point. Fit coefficients in direction , , and Fit coefficients in direction , , .

[0122] Among them, quadratic polynomial fitting describes the motion trend of historical trajectories by establishing a quadratic function relationship between coordinates and time; the latest historical trajectory... These points are continuous at the end of the historical trajectory. A set of location points.

[0123] (2) Based on the fitting coefficient For the new trajectory set time domain Perform location prediction and generate predicted location sequences. .

[0124] For example, determining a new trajectory set The time domain range is defined as the time interval of the new trajectory. Substitute each time value within this time domain sequentially into... Quadratic polynomial in the direction (including fitting coefficients) , , )and Quadratic polynomial in the direction (including fitting coefficients) , , In ), the corresponding calculations are obtained respectively. Predicted coordinates and Predict coordinates, all of which are Predicted coordinates and The predicted location points, composed of predicted coordinates, are arranged in chronological order to generate a predicted location sequence.

[0125] The time domain of the new trajectory set is the time range from the start to the end of the new trajectory; the predicted position sequence is a set composed of multiple predicted position points in chronological order.

[0126] In one embodiment, the distance loss matrix and the feature parameter loss matrix are weighted and fused to generate an overall association cost matrix, including:

[0127] (1) For each historical trajectory in the coarse correlation trajectory pair set Calculate the interruption time length based on the new trajectory j. ,in For the historical trajectory The final moment, This marks the starting point of the new trajectory.

[0128] For example, historical trajectory The final moment is the trajectory of history. The time stamp of the end position; the starting time of the new trajectory j is the time stamp of the starting position of the new trajectory j.

[0129] (2) Based on the preset attenuation factor and interruption duration Preset attenuation factor Based on experimental data, the dynamic weighting coefficients were calculated. ;

[0130] Calculate complementary weighting coefficients .

[0131] For example, a preset attenuation factor Used to control the decay rate of weights as the interruption time changes; dynamic weight coefficient Reflecting the dynamic change of the weighting of distance loss in fusion with interruption time; complementary weighting coefficients It reflects the weight ratio of feature parameter loss in the fusion process.

[0132] (3) For the distance loss matrix Middle elements With feature parameter loss matrix Middle elements Perform weighted fusion:

[0133]

[0134] Iterate through all coarsely correlated trajectory pairs and fuse the results of each trajectory pair. Fill in the matrix to generate the overall association cost matrix. .

[0135] For example, the distance loss matrix The matrix is ​​formed by arranging the average distance loss values ​​of each coarsely correlated trajectory pair in rows and columns; the feature parameter loss matrix. The overall association cost matrix is ​​a matrix formed by arranging the feature parameter loss values ​​of each coarse correlation trajectory pair in rows and columns. It is the core matrix that integrates distance and feature parameter loss information for subsequent trajectory association decisions.

[0136] In one embodiment, the historical trajectories in the coarse correlation trajectory pair set are... The latest Points Perform a quadratic polynomial fitting and solve for the fitting coefficients. ,include:

[0137] (1) Obtain historical trajectories from the coarse correlation trajectory pair set The latest A sequence of positions of points, where the position information of each point is... , For the historical trajectory Total points.

[0138] For example, for historical trajectories in the coarse correlation trajectory pair set Determine the historical trajectory The total number of points is From the historical trajectory Select the end continuous from all position points Individual points form the historical trajectory. The latest The sequence of positions of points. The range of each point is the historical trajectory. The Point 1 to the 2nd There are points, each containing _ points. coordinate , coordinate and time And all points are sorted in ascending order of time.

[0139] Among them, historical trajectory Total points It is the trajectory of history Total number of location points included; historical trajectory The latest The position sequence of each point is a historical trajectory. end The set of location points arranged in chronological order is used for subsequent fitting analysis.

[0140] (2) For the position sequence The relationship between coordinates and time is fitted using a quadratic polynomial to obtain... Directional fitting polynomial: ,in for Directional fitting coefficients;

[0141] For position sequence The relationship between coordinates and time is fitted using a quadratic polynomial to obtain... Directional fitting polynomial: ,in for Directional fitting coefficients;

[0142] based on Directional fitting coefficients and Directional fitting coefficients, obtaining the fitting coefficients .

[0143] For example, regarding historical trajectories The latest The sequence of positions of points, with time as the independent variable. With coordinates as the dependent variable, construct... Quadratic polynomial model in the direction (in the form of ), calculate the actual value of each point Coordinates and Model Prediction The coordinate error is obtained by minimizing the sum of the squares of all errors. Fit coefficients in direction , , Using time as the independent variable, With coordinates as the dependent variable, construct... Quadratic polynomial model in the direction (in the form of ), calculate the actual value of each point Coordinates and Model Prediction The coordinate error is obtained by minimizing the sum of the squares of all errors. Fit coefficients in direction , , .Will direction and By integrating the fitting coefficients of the directions, we obtain the fitting coefficients. .

[0144] in, The quadratic polynomial model of the direction is used to describe The pattern of coordinate changes over time; The quadratic polynomial model of the direction is used to describe The coordinates change over time; the sum of squared errors is the sum of the squares of the differences between the actual and predicted values; minimizing this value makes the model fit the actual trajectory better; fitting coefficients. These parameters determine the quadratic polynomial model and are used for subsequent trajectory prediction.

[0145] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0146] In one embodiment, such as Figure 2 As shown, this application also provides a target interruption trajectory association system 200 based on a fusion strategy, the system 200 comprising:

[0147] The coarse correlation prediction module 201 is used to perform distance constraint filtering on the historical trajectory set and the new trajectory set to obtain a coarse correlation trajectory pair set; and to perform trajectory prediction processing on the historical trajectories in the coarse correlation trajectory pair set to generate a predicted position sequence.

[0148] The loss feature construction module 202 is used to calculate the average distance between the predicted position sequence and the position sequence of the new trajectory set, obtain the average distance value, and construct the distance loss matrix based on the average distance value; and to extract feature parameters from the historical trajectories and new trajectories in the coarse correlation trajectory pair set to obtain velocity features, acceleration features and angular velocity features.

[0149] The fusion decision module 203 is used to calculate the feature parameter loss of velocity features, acceleration features and angular velocity features, and generate the feature parameter loss matrix; to perform weighted fusion processing on the distance loss matrix and the feature parameter loss matrix to generate the overall association cost matrix; and to perform Hungarian algorithm optimization processing on the overall association cost matrix to output the associated trajectory pairs.

[0150] Specifically, the coarse correlation prediction module 201 performs distance constraint filtering on the historical trajectory set and the new trajectory set. By calculating the distance between the end point of the historical trajectory and the starting point of the new trajectory, and combining this with the target maximum speed to determine the feasibility of the motion, a set of coarse correlation trajectory pairs is obtained. For each historical trajectory in the set of coarse correlation trajectory pairs, its latest position point sequence is selected for fitting to establish a trajectory prediction model. Based on this model, the position of the new trajectory set in the time domain is predicted, generating a predicted position sequence.

[0151] The historical trajectory set contains multiple historical trajectories, and the new trajectory set contains multiple new trajectories. Distance constraint filtering is a preliminary correlation filtering based on the distance between trajectory endpoints and the feasibility of movement. Trajectory prediction predicts the position of the trajectory in the time domain of the new trajectory by fitting the motion patterns of historical trajectories; the predicted position sequence is a set of predicted position points arranged in chronological order.

[0152] The loss feature construction module 202 calculates the average distance for each pair of predicted position sequences and new trajectory sets, and arranges these average distance values ​​according to the correspondence between historical and new trajectories to construct a distance loss matrix. For each historical and new trajectory in the coarsely correlated trajectory pair set, its velocity, acceleration, and angular velocity features are extracted. For historical trajectories, the end position point sequence is selected; for new trajectories, the starting position point sequence is selected. Through steps such as fitting, synthesis, and calculation of the motion velocity and direction angle of adjacent points, three types of features for the two types of trajectories are obtained.

[0153] The average distance calculation involves averaging the distances between the predicted position and the new trajectory position; the distance loss matrix is ​​a matrix that quantifies the distance differences between trajectories. Velocity features are the quantified results of trajectory motion velocity, acceleration features are the quantified results of velocity change rate, and angular velocity features are the quantified results of direction angle change rate.

[0154] The fusion decision module 203 calculates the degree of feature difference between historical and new trajectories for velocity, acceleration, and angular velocity features. After normalization and weight integration, the feature parameter loss values ​​are arranged according to the correspondence between historical and new trajectories to generate a feature parameter loss matrix. The interruption time length of coarsely associated trajectory pairs is calculated. Dynamic weight coefficients and complementary weight coefficients are determined based on a preset attenuation factor. The corresponding elements of the distance loss matrix and feature parameter loss matrix are weighted and summed to fill the matrix and generate the overall association cost matrix. A function is constructed with the objective of minimizing the total association cost, setting constraints that each historical trajectory can be associated with at most one new trajectory and each new trajectory can only be associated with one historical trajectory. The allocation matrix is ​​solved using the Hungarian algorithm. Based on the elements identifying association in the allocation matrix, associated trajectory pairs are output.

[0155] The feature parameter loss matrix quantifies the feature differences between trajectories. The overall association cost matrix is ​​a decision matrix that fuses distance and feature difference information. The Hungarian algorithm is used to solve for the optimal trajectory association assignment; associated trajectory pairs are combinations of historical trajectories and new trajectories that are determined to have an association relationship.

[0156] The coarse correlation prediction module 201 is also used for:

[0157] Obtain historical trajectory set and new trajectory set ,in , Representing historical trajectory The A location point, Indicates new trajectory The A location point, Representing historical trajectory The Location points coordinate, Representing historical trajectory The Location points coordinate, Representing historical trajectory The Location points coordinate;

[0158] Historical trajectory set The finish line With the new trajectory set starting position Perform Euclidean distance calculation to obtain the distance value. ,in, Representing historical trajectory The timestamp of the endpoint location, Representing historical trajectory End point coordinate, Representing historical trajectory End point coordinate, Indicates new trajectory starting point coordinate, Indicates new trajectory starting point coordinate, Indicates new trajectory The timestamp of the starting point;

[0159] Based on the preset target speed For distance value Perform a feasibility assessment of the motion and output that the motion is satisfied. The set of coarse correlation trajectories, where, Indicates the distance value. Indicates the target's maximum speed. Indicates the starting time of the new trajectory. It indicates the endpoint of the historical trajectory.

[0160] The fusion decision module 203 is also used for:

[0161] Construct the objective function: ,in For the number of historical trajectories, For the number of new trajectories, These are the elements of the overall association cost matrix. Represent decision variables;

[0162] Set constraints:

[0163]

[0164] and ;

[0165] Solve for the assignment matrix that satisfies the constraints. ;

[0166] According to the allocation matrix China satisfies The elements output historical trajectory With new trajectory The associated trajectory pairs.

[0167] Loss feature construction module 202 is also used for:

[0168] On the trajectory of history The end The historical trajectory is obtained by performing least-squares fitting on a sequence of location points. Directional velocity components and Directional velocity components ;

[0169] right and The velocity characteristics of the historical trajectories were obtained through synthetic processing. ;

[0170] The final historical trajectory Calculate the velocity of adjacent points from a sequence of location points to generate a velocity sequence of historical trajectories. ;

[0171] For velocity sequences Least squares fitting was performed to obtain the acceleration characteristics of the historical trajectory. ;

[0172] The final historical trajectory Calculate the orientation angles of adjacent points from a sequence of location points to generate an orientation angle sequence of the historical trajectory. ;

[0173] Opposite angle sequence Least squares fitting was performed to obtain the angular velocity characteristics of the historical trajectory. ;

[0174] New trajectory The front The sequence of position points is fitted using the least squares method to obtain the new trajectory. Directional velocity components and Directional velocity components ;

[0175] right and The velocity characteristics of the new trajectory are obtained through synthesis processing. ;

[0176] Foremost of the new trajectory Calculate the velocity sequence of adjacent points from a sequence of location points to generate a velocity sequence for a new trajectory. ;

[0177] For velocity sequences The acceleration characteristics of the new trajectory are obtained by performing least squares fitting. ;

[0178] Foremost of the new trajectory Calculate the orientation angles of adjacent points from a sequence of location points to generate a new trajectory orientation angle sequence. ;

[0179] Opposite angle sequence The angular velocity characteristics of the new trajectory are obtained by performing least squares fitting. ;

[0180] Velocity characteristics include the velocity characteristics of historical trajectories. and the velocity characteristics of the new trajectory Acceleration characteristics include acceleration characteristics of historical trajectories. and acceleration characteristics of the new trajectory Angular velocity characteristics include the angular velocity characteristics of historical trajectories. Angular velocity characteristics of the new trajectory .

[0181] The coarse correlation prediction module 201 is also used for:

[0182] Historical trajectories in the coarse correlation trajectory pair set The latest Points Perform a quadratic polynomial fitting and solve for the fitting coefficients. ;

[0183] Based on fitting coefficients For the new trajectory set time domain Perform location prediction and generate predicted location sequences. .

[0184] The fusion decision module 203 is also used for:

[0185] For each historical trajectory in the coarse correlation trajectory pair set Calculate the interruption time length based on the new trajectory j. ,in For the historical trajectory The final moment, This marks the starting point of the new trajectory.

[0186] Based on preset attenuation factor and interruption duration Preset attenuation factor Based on experimental data, the dynamic weighting coefficients were calculated. ;

[0187] Calculate complementary weighting coefficients ;

[0188] For distance loss matrix Middle elements With feature parameter loss matrix Middle elements Perform weighted fusion:

[0189]

[0190] Iterate through all coarsely correlated trajectory pairs and fuse the results of each trajectory pair. Fill in the matrix to generate the overall association cost matrix. .

[0191] The coarse correlation prediction module 201 is also used for:

[0192] Obtain historical trajectories from the coarse correlation trajectory pair set The latest A sequence of positions of points, where the position information of each point is... , For the historical trajectory Total points;

[0193] For position sequence The relationship between coordinates and time is fitted using a quadratic polynomial to obtain... Directional fitting polynomial: ,in for Directional fitting coefficients;

[0194] For position sequence The relationship between coordinates and time is fitted using a quadratic polynomial to obtain... Directional fitting polynomial: ,in for Directional fitting coefficients;

[0195] based on Directional fitting coefficients and Directional fitting coefficients, obtaining the fitting coefficients .

[0196] In one embodiment, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0197] In one embodiment, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0198] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0199] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for associating target interrupt trajectories based on fusion strategy, characterized in that, The method includes: Distance constraint filtering is performed on the historical trajectory set and the new trajectory set to obtain a set of coarsely correlated trajectory pairs; The historical trajectories in the coarse correlation trajectory set are subjected to trajectory prediction processing to generate a predicted position sequence; The average distance between the predicted position sequence and the position sequence of the new trajectory set is calculated to obtain the average distance value, and a distance loss matrix is ​​constructed based on the average distance value; feature parameters are extracted from the historical trajectories and new trajectories in the coarse correlation trajectory set to obtain velocity features, acceleration features, and angular velocity features. The velocity feature, acceleration feature, and angular velocity feature are subjected to feature parameter loss calculation to generate a feature parameter loss matrix; the distance loss matrix and the feature parameter loss matrix are subjected to weighted fusion processing to generate an overall association cost matrix; the overall association cost matrix is ​​optimized by the Hungarian algorithm to output associated trajectory pairs.

2. The method of claim 1, wherein, The distance constraint filtering process performed on the historical trajectory set and the new trajectory set yields a coarsely correlated trajectory pair set, including: set of historical trajectories and a set of new trajectories wherein , denotes a th position point of a historical trajectory , denotes a th position point of a new trajectory , denotes a th coordinate of a th position point of a historical trajectory , denotes a th coordinate of a th position point of a historical trajectory , denotes a th coordinate of a th position point of a historical trajectory ; end point position of the historical trajectory set end point position of the historical trajectory set start point position of the new trajectory set start point position of the new trajectory set Euclidean distance calculation is performed on the start point position of the new trajectory set wherein, the timestamp of the end point position point of the historical trajectory the coordinate of the end point position point of the historical trajectory the coordinate of the end point position point of the historical trajectory the coordinate of the end point position point of the historical trajectory the coordinate of the start point position point of the new trajectory the coordinate of the start point position point of the new trajectory the coordinate of the start point position point of the new trajectory the timestamp of the start point position point of the new trajectory the timestamp of the start point position point of the new trajectory the timestamp of the start point position point of the new trajectory the timestamp of the start point position point of the new trajectory the timestamp of the start point position point of the new trajectory the timestamp of the start point position point of the new trajectory the timestamp of the start point position point of the new trajectory the timestamp of the start point position point of the new trajectory ​ Based on a preset target speed On the distance value Motion feasibility is judged, and the coarse association trajectory pair set satisfying , is output, wherein, Indicates the distance value, Indicates the target maximum speed, Indicates the starting time of the new trajectory, Indicates the end time of the historical trajectory.

3. The method of claim 1, wherein, The process of optimizing the overall association cost matrix using the Hungarian algorithm to output association trajectory pairs includes: Constructing the objective function: where is the number of historical trajectories, is the number of new trajectories, is an element of the overall cost matrix, denotes the decision variable; Setting constraints: and ; solving for a distribution matrix that satisfies the constraints ; According to the assignment matrix satisfying the element, the output history trajectory and the associated trajectory pair of the new trajectory .

4. The method of claim 1, wherein, The step of extracting feature parameters from the historical and new trajectories in the coarse correlation trajectory set to obtain velocity features, acceleration features, and angular velocity features includes: On the trajectory of history The end The historical trajectory is obtained by performing least-squares fitting on a sequence of location points. Directional velocity components and Directional velocity components ; Synthesis processing is performed on the historical trajectory speed feature and to obtain a historical trajectory speed feature ; calculating a motion velocity of adjacent points for the last position point sequence of the historical trajectory, to generate a velocity sequence of the historical trajectory ; performing a least square fitting process on the velocity sequence to obtain an acceleration feature of the historical trajectory ; calculating a direction angle of adjacent points for the last position point sequence of the historical trajectory, to generate a direction angle sequence of the historical trajectory ; performing a least square fitting process on the sequence of direction angles to obtain an angular velocity feature of the historical trajectory ; For the new trajectory The front The sequence of position points is fitted using the least squares method to obtain the new trajectory. Directional velocity components and Directional velocity components ; To and synthesis processing, get the speed characteristics of the new track ; calculating a motion speed of adjacent points for the sequence of the new trajectory's first position points to generate a speed sequence of the new trajectory ; performing a least square fitting process on the velocity sequence to obtain acceleration characteristics of the new trajectory ; calculating a direction angle of adjacent points for the sequence of the new track's first position points to generate a sequence of direction angles of the new track ; performing a least square fitting process on the sequence of direction angles to obtain angular velocity characteristics of a new trajectory ; the speed feature of the new trajectory and the speed feature of the new trajectory the acceleration feature of the new trajectory and the acceleration feature of the new trajectory the angular velocity feature of the new trajectory and the angular velocity feature of the new trajectory .

5. The method of claim 1, wherein, The step of performing trajectory prediction processing on historical trajectories in the coarse correlation trajectory pair set to generate a predicted position sequence includes: For the historical trajectories in the coarse correlation trajectory pair set The latest Points Perform a quadratic polynomial fitting and solve for the fitting coefficients. ; based on the fitting coefficients for the new trajectory set time domain perform position prediction to generate the predicted position sequence .

6. The method of claim 1, wherein, The weighted fusion process of the distance loss matrix and the feature parameter loss matrix to generate the overall association cost matrix includes: For each historical trajectory in the set of coarse correlation trajectory pairs Calculate the interruption time length based on the new trajectory j. ,in For the historical trajectory The final moment, This marks the starting point of the new trajectory. based on a preset attenuation factor and the interrupt time length , the preset attenuation factor based on experimental data, calculate dynamic weight coefficient ; Computing complementary weight coefficients ; elements in the distance loss matrix elements in the distance loss matrix elements in the feature parameter loss matrix elements in the feature parameter loss matrix weighted fusion: traversing all coarse association track pairs, fusing results of each track pair filling in the matrix, generating the global association cost matrix .

7. The method of claim 5, wherein the method further comprises: The latest point of the historical trajectory in the coarse association trajectory pair set , comprising:​​​ acquire a position sequence of the latest points of each historical trajectory in the coarse association trajectory pair set, wherein the position information of each point is , , the total number of points of the historical trajectory ​ for the position sequence The relationship between the coordinates and time is fitted by a quadratic polynomial to obtain The direction fitting polynomial is: wherein is the direction fitting coefficient the direction fitting coefficient to the position sequence The relationship between the coordinates and time is fitted by a quadratic polynomial to obtain The direction fitting polynomial is: wherein is the direction fitting coefficient; based on the direction fitting coefficients and the direction fitting coefficients, resulting in the fitting coefficients .

8. A target interrupt trajectory association system based on a fusion strategy, characterized in that, The system includes: The coarse correlation prediction module is used to perform distance constraint filtering on the historical trajectory set and the new trajectory set to obtain a set of coarse correlation trajectory pairs; The historical trajectories in the coarse correlation trajectory set are subjected to trajectory prediction processing to generate a predicted position sequence; The loss feature construction module is used to calculate the average distance between the predicted position sequence and the position sequence of the new trajectory set to obtain the average distance value, and construct a distance loss matrix based on the average distance value; and to perform feature parameter extraction processing on the historical trajectories and new trajectories in the coarse correlation trajectory pair set to obtain velocity features, acceleration features and angular velocity features. The fusion decision module is used to calculate the feature parameter loss of the velocity feature, the acceleration feature, and the angular velocity feature to generate a feature parameter loss matrix; to perform weighted fusion processing on the distance loss matrix and the feature parameter loss matrix to generate an overall association cost matrix; and to perform Hungarian algorithm optimization processing on the overall association cost matrix to output associated trajectory pairs. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. When the processor executes the computer program, it implements the steps of the target interrupt trajectory association method based on the fusion strategy according to any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the target interruption trajectory association method based on the fusion strategy according to any one of claims 1 to 7.

Images