Interrupted Track Association Method, System, Terminal and Medium Based on Track Prediction

Abstract

The present invention belongs to the field of radar target tracking, and specifically relates to a method, system, terminal and medium for interrupted track association based on track prediction, which roughly associates old tracks and new tracks, assigns values to the association matrix according to the rough association results; normalizes the track data of all track pairs after rough association; based on the preprocessed old track data, obtains the track data of the predicted new track through a pre-trained track prediction model based on time series; based on the preprocessed old track data, new track data and predicted new track data, obtains the association classification result of the track pair through a pre-trained time series classification model; reassigns values to the association matrix according to the association classification result, and obtains the interrupted track association result based on the reassigned association matrix. The present invention improves the accuracy of interrupted track association under a long interruption duration through time series classification and time series prediction.

Description

Technical Field

[0001] The present invention belongs to the field of radar target tracking, and particularly relates to a method, a system, a terminal and a medium for interrupted track association based on track prediction. Background Art

[0002] Due to reasons such as platform movement, target maneuver, and long radar sampling intervals, track interruption often occurs, which in turn causes serious interference to subsequent target tracking, situation awareness, information fusion, etc. Therefore, interrupted track association has become one of the key problems that need to be solved urgently in the field of radar data processing. The traditional TSA (Time Series Analysis) method first needs to obtain prior knowledge, assume a target motion model, and perform complex calculations on the interrupted track data, compare the differences between each track, and finally complete the association task through association assignment. However, when performing track prediction, there are often unreasonable assumptions, especially in long-term interruption and dense association environments, and the association effect is often unsatisfactory.

[0003] Currently, related technologies apply deep learning methods to the problem of interrupted track association. For example, a neural network model including a track feature extraction module, a time comparison module, a track feature comparison module, and a classifier module is used to calculate the association probability of two tracks. However, this method is applicable to short interruption durations and can achieve good association effects in the case of short interruption durations, but the association effect for interrupted tracks with long interruption durations is not ideal. Summary of the Invention

[0004] To solve the above problems, the present invention provides a method, a system, a terminal and a medium for interrupted track association based on track prediction, which uses a track prediction model based on time series for track prediction, and uses a time series classification model for association classification, and performs interrupted track association through the combination of time series classification and time series prediction, so as to improve the accuracy of interrupted track association under long interruption durations.

[0005] In a first aspect, the technical solution of the present invention provides a method for interrupted track association based on track prediction, including the following steps:

[0006] Coarsely associate all old tracks and new tracks in the scene based on time matching and speed matching, and assign values to the association matrix according to the coarse association results;

[0007] Normalize the track data of all track pairs after coarse association;

[0008] Preprocess the track data of old tracks and new tracks, and the preprocessing includes mean filtering and variable dimension expansion operations;

[0009] Based on the preprocessed old track data, track data for predicting new tracks is obtained through a pre-trained track prediction model based on time series;

[0010] Based on the preprocessed old track data, new track data, and predicted new track data, an associated classification result of the track pair is obtained through a pre-trained time series classification model;

[0011] Reassign the association matrix according to the associated classification result, and obtain the interrupted track association result based on the reassigned association matrix.

[0012] In an optional implementation, based on the preprocessed old track data, obtaining track data for predicting new tracks through a pre-trained track prediction model based on time series specifically includes:

[0013] Perform dimensional transformation on the input data through the first linear layer; the input data refers to the preprocessed old track data;

[0014] The output data of the first linear layer flows into the Mamba model and the reverse Mamba model respectively, and processes the sequence data from the forward and reverse directions respectively;

[0015] Combine the output data of the Mamba model and the reverse Mamba model through an addition operation;

[0016] Normalize the combined data through the first normalization layer;

[0017] Extract non-linear features from the output data of the first normalization layer through the feed-forward network layer;

[0018] Perform an addition operation on the output data of the first normalization layer and the feed-forward network layer and then input it into the second normalization layer for normalization processing;

[0019] Input the output data of the second normalization layer into the second linear layer, and the second linear layer outputs the track data for predicting new tracks.

[0020] In an optional implementation, the objective function of the track prediction model in the training stage is to use contrastive loss to calculate the loss value between the predicted new track and the actual new track's track data, defined as:

[0021]

[0022] Among them, is a binary label. If the two track segments of the old track and the actual new track in the sample come from the same target, then otherwise ; Denote the interval distance, which is the shortest distance between two track segments belonging to different targets in the sample; Denote the Euclidean distance between the predicted new track and the actual new track.

[0023] In an alternative embodiment, based on the preprocessed old track data, new track data, and predicted new track data, obtain the association classification result of the track pair through a pre-trained time series classification model, specifically including:

[0024] Perform a subtraction operation on the preprocessed old track data and new track data;

[0025] Perform a concatenation operation on the subtraction operation result, new track data, and predicted new track data;

[0026] Input the concatenation result into the pre-trained time series classification model, and the time series classification model outputs the association classification result of the track pair.

[0027] In an alternative embodiment, the objective function of the time series classification model in the training phase is to use the cross-entropy loss function to calculate the loss value between the predicted probability and the true distribution. The output of the time series classification model for each sample is , is the probability that the old track and the new track do not belong to the same target, is the probability that the old track and the new track belong to the same target. The true label is 0 or 1, and the corresponding cross-entropy loss function is expressed as:

[0028]

[0029] where is the number of samples;

[0030] The total loss function is expressed as: .

[0031] In an alternative embodiment, input the concatenation result into the pre-trained time series classification model, and the time series classification model outputs the association classification result of the track pair, specifically including:

[0032] Let the concatenation result flow into the increasing filter, decreasing filter, peak filter, and the first Inception module respectively;

[0033] Perform a concatenation operation on the output data of the increasing filter, decreasing filter, peak filter, and the first Inception module to obtain the integrated feature;

[0034] The integrated feature is extracted by the second Inception module twice to obtain multi-scale features;

[0035] Perform the first residual connection on the input data and the extracted multi-scale features;

[0036] The result of the first residual connection is passed through the third Inception module three times to extract multi-scale features again;

[0037] Perform the second residual connection on the result of the second residual connection and the multi-scale features extracted again;

[0038] Reduce the feature dimension of the result of the second residual connection through the global average pooling layer;

[0039] Output the association classification result of the track pair through the linear layer for the output data of the global average pooling layer.

[0040] In an alternative embodiment, reassign the association matrix according to the association classification result, and obtain the interrupted track association result based on the reassigned association matrix, specifically including:

[0041] The association classification result includes the th old track and the th new track The probability of not belonging to the same target and the probability of belonging to the same target ;

[0042] The element in the th row and the th column of the association matrix is the th old track and the th new track The association flag ; Indicates that the two are not associated, Indicates that the two are associated;

[0043] When and the current association flag , update the association flag to 0;

[0044] Define the set and initialize it to be empty, which is used to store the sequence numbers of the newly and old tracks that are finally associated;

[0045] Traverse the track pairs in the association matrix that satisfy , find the and the sequence number of the old track with the largest value in each group of track pairs and the new track , and mark it as ;

[0046] Mark Added to the set and set all elements in the th row and the th column of the association matrix to 0, and then traverse the track pairs in the association matrix that satisfy again; finally, the content in the set is the interrupted track association result.

[0047] In a second aspect, the technical solution of the present invention provides an interrupted track association system based on track prediction, including:

[0048] A track rough association unit for roughly associating all old tracks and new tracks in the scene based on time matching and speed matching, and assigning values to the association matrix according to the rough association result;

[0049] A track data normalization processing unit for normalizing the track data of all track pairs after rough association;

[0050] A track data preprocessing unit for preprocessing the track data of old tracks and new tracks, and the preprocessing includes mean filtering and variable dimension expansion operations;

[0051] A track data prediction unit for obtaining the track data of the predicted new track based on the preprocessed old track data through a pre-trained track prediction model based on time series;

[0052] An association classification unit for obtaining the association classification result of the track pair based on the preprocessed old track data, new track data, and predicted new track data through a pre-trained time series classification model;

[0053] An association result acquisition unit for reassigning values to the association matrix according to the association classification result to obtain the interrupted track association result.

[0054] In a third aspect, the technical solution of the present invention provides a terminal, including:

[0055] A memory for storing an interrupted track association program based on track prediction;

[0056] A processor for implementing the steps of the interrupted track association method based on track prediction as described in any one of the above when executing the interrupted track association program based on track prediction.

[0057] In a fourth aspect, the technical solution of the present invention provides a computer-readable storage medium, on which an interrupted track association program based on track prediction is stored, and when the interrupted track association program based on track prediction is executed by a processor, the steps of the interrupted track association method based on track prediction as described in any one of the above are implemented.

[0058] As can be seen from the above technical solutions, the present application has the following advantages: First, the new and old tracks are roughly correlated, and the track data of the track segments are normalized and preprocessed. Then, through a track prediction model based on time series, the old track is used to predict the new track. Next, according to the prediction results and the track data of the actual track segments, correlation classification is performed through a time series classification model. Finally, the interrupted track association is obtained based on the classification results. The present invention uses a track prediction model based on time series for track prediction, which can fully learn the time series features and motion laws in the old track. The prediction method based on historical data not only considers the motion inertia of the target but also can capture the trend changes of the target's motion. Through in-depth analysis of the old track, the prediction model can more accurately predict the possible direction and state of the new track. And a time series classification model is used for correlation classification, which can comprehensively analyze various factors such as the similarity, difference, and matching degree of time series features between the predicted track and the actual track, and then accurately judge whether there is an association relationship between track pairs, avoiding association errors caused by unreasonable simple assumptions of the target motion model, and being able to more flexibly adapt to different target motion patterns and complex environmental conditions, thus significantly improving the accuracy of track association judgment. In the case of a long interruption duration, the time series prediction model uses the historical information of the old track to accurately infer the motion track of the target during the interruption; the time series classification model can analyze the association relationship between tracks based on the prediction results and the actual track data. The two cooperate with each other, so that in the scenario of a long interruption duration, a high track association accuracy can still be maintained, and the accuracy of interrupted track association in the case of a long interruption duration is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions of the present application, the accompanying drawings required for description will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0060] Figure 1 Schematic flow chart of a method for interrupted track association based on track prediction provided by an embodiment of the present invention.

[0061] Figure 2 Schematic diagram of the framework structure for interrupted track association based on track prediction.

[0062] Figure 3 Schematic diagram of track data change.

[0063] Figure 4 Schematic diagram of the S-mamba model architecture.

[0064] Figure 5It is a schematic diagram of the Mamba model architecture.

[0065] Figure 6 It is a schematic diagram of the H-Inception Time model architecture.

[0066] Figure 7 It is Figure 6 a schematic diagram of the filter structure manually created in

[0067] Figure 8 It is Figure 6 a schematic diagram of the Inception module structure in

[0068] Figure 9 a schematic diagram of the complete track of 5000 targets containing noise.

[0069] Figure 10 It is a schematic block diagram of an interrupted track association system based on track prediction provided by an embodiment of the present invention.

[0070] Figure 11 It is a schematic diagram of the structure of a terminal provided by an embodiment of the present invention. Detailed implementation manners

[0071] To make the application purpose, features, and advantages of the present application more obvious and understandable, the technical solutions protected by the present application will be clearly and completely described below by using specific embodiments and accompanying drawings. Obviously, the embodiments described below are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in this patent, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this patent.

[0072] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this invention belongs. The terms used in the specification of this invention herein are only for the purpose of describing specific embodiments, and are not intended to limit this invention.

[0073] Inception module: It is composed of multiple parallel convolutional layers of different sizes, can capture information at different spatial scales, enable the model to learn rich representations of the input data, and effectively capture local and global patterns in the data.

[0074] Mamba model: It is a new type of State Space Model (SSM), mainly developed by Albert Gu (Carnegie Mellon University) and Tri Dao (Princeton University).

[0075] Figure 1Schematic flowchart of an interrupted track association method based on track prediction provided by an embodiment of the present invention. Among them, Figure 1 The execution subject can be a system for interrupted track association based on track prediction. The interrupted track association method provided by the embodiment of the present invention is executed by a computer device. Correspondingly, the system for interrupted track association based on track prediction runs in the computer device. According to different requirements, the order of steps in this flowchart can be changed, and some can be omitted.

[0076] As Figure 1 shown, the method includes the following steps.

[0077] S1. Coarsely associate all old tracks and new tracks in the scene based on time matching and speed matching, and assign values to the association matrix according to the coarse association results.

[0078] S2. Normalize the track data of all track pairs after coarse association.

[0079] S3. Preprocess the track data of old tracks and new tracks. The preprocessing includes mean filtering and variable dimension expansion operations.

[0080] S4. Based on the preprocessed old track data, obtain the track data of the predicted new track through a pre-trained track prediction model based on time series.

[0081] S5. Based on the preprocessed old track data, new track data, and predicted new track data, obtain the association classification result of the track pair through a pre-trained time series classification model.

[0082] S6. Reassign values to the association matrix according to the association classification result, and obtain the interrupted track association result based on the reassigned association matrix.

[0083] In this embodiment, a trajectory prediction model based on time series is adopted for trajectory prediction, which can fully learn the time series features and motion laws in the old trajectory. The prediction method based on historical data not only considers the motion inertia of the target but also can capture the trend changes of the target's motion. Through in-depth analysis of the old trajectory, the prediction model can relatively accurately predict the possible directions and states of the new trajectory. And a time series classification model is used for correlation classification, which can comprehensively analyze various factors such as the similarity, difference, and matching degree of time series features between the predicted trajectory and the actual trajectory, and then accurately judge whether there is a correlation relationship between track pairs, avoiding correlation errors caused by unreasonable simple assumptions of the target motion model, and being able to more flexibly adapt to different target motion patterns and complex environmental conditions, thus significantly improving the accuracy of track correlation judgment. In the case of a long interruption duration, the time series prediction model uses the historical information of the old trajectory to accurately infer the motion trajectory of the target during the interruption; the time series classification model can analyze the correlation relationship between tracks based on the prediction results and the actual track data. The two cooperate with each other, so that in the scenario of a long interruption duration, a high track correlation accuracy can still be maintained, and the correlation accuracy of interrupted tracks under a long interruption duration is improved.

[0084] Further, as a refinement and extension of the specific implementation manner of the above embodiment, in order to fully illustrate the specific implementation process in this embodiment, another method for associating interrupted tracks based on trajectory prediction is provided. This method first filters out unlikely associated old and new tracks through rough association, performs mean filtering on the remaining old and new tracks to make the track data "smooth", and at the same time performs first-order difference for variable dimension expansion. The result after differentiation is concatenated with the previous data to increase the variable dimension. Then, based on the Mamba model, the old track is used to predict the new track, and the prediction result is compared with the new track, so that the distance between the determined associated old and new tracks becomes smaller, and the distance between the determined unassociated old and new tracks becomes larger. Finally, the prediction result, the new track, and the difference between the old and new tracks before prediction are concatenated, and the concatenated result is classified by the H-Inception model to obtain the probability of whether the old and new tracks are associated. This method includes the following steps.

[0085] SS1, perform rough association on all old and new tracks in the scenario based on time matching and speed matching, and assign values to the association matrix according to the rough association result.

[0086] When a target has a trajectory interruption, multiple trajectory segments may be generated for the same target. In a scenario of associating interrupted tracks, there may be a phenomenon where multiple targets have trajectory interruptions. The task of associating interrupted tracks is to pair and associate the temporally adjacent trajectory segments belonging to the same target in this scenario. In the original trajectory data, each trajectory contains information in multiple dimensions, where the Location information, Position information, speed scalar information and corresponding time information are used as characteristic information associated with tracks, namely track data. The time corresponding to the first point track sampled for a track segment is defined as the start time of the track segment, and the time corresponding to the last point track is defined as the end time of the track segment. When the end time of a track segment is before the start time of another track segment, the former is defined as the old track and the latter is defined as the new track. A pair of old tracks and new tracks that need to be judged whether they are associated is called a track pair. The start and end times of all track segments in the scene are counted to obtain the maximum start time of all track segments. and the minimum end time is The definitions of new and old tracks are as follows:

[0087] Old track: The end time of the track segment is less than track segment.

[0088] (1)

[0089] New track: The start time of the track segment is greater than track segment.

[0090] (2)

[0091] in Indicates An old track, Indicates New track, Indicates Old track Status information at all times, Indicates New tracks Status information at all times, and Respectively represent The start and end times of the old track, and Respectively represent The start and end times of the new track, and Respectively represent the number of old and new tracks.

[0092] definition OK Column Correlation Matrix .

[0093] (3)

[0094] In the formula Indicates the judgment of the th old track and the th new track are associated. Indicates the judgment of the th old track and the th new track are not associated. After calculating through the interrupted track association algorithm introduced in this embodiment, each old track is associated with at most one new track, and each new track is associated with at most one old track, that is, at most one element in each row (each column) of the association matrix is 1, and the rest are 0.

[0095] Only discuss the situation where the target is in a two-dimensional rectangular coordinate system. When the old track and the new track are the front and back two track segments interrupted on the same track of the same target, then and must satisfy the matching relationship in terms of time and speed, that is, the start time of the new track is greater than the end time of the old track , and the ratio of the interruption distance to the interruption time is less than or equal to the maximum moving speed of the target. In order to reduce the association time, avoid the model over-focusing on irrelevant samples, and at the same time exclude most of the interrupted track segments that are unlikely to be associated, the initial association judgment is made on all old tracks and new tracks in the scene based on time matching and speed matching respectively. Based on the rough association result, the association matrix is assigned. If the old track and the new track simultaneously satisfy time matching and speed matching, then let the old track and the new track correspond to in the association matrix , otherwise let .

[0096] (4)

[0097] Equation (4) represents the track pair that simultaneously satisfies time matching and speed matching. and respectively represent the positions of the end time of the th old track in the direction and the direction. and respectively represent the positions of the start time of the th new track in the direction and the The position in the .

[0098] SS2, normalize the track data of all track pairs after rough association.

[0099] To avoid numerical problems and enhance the stability of the model, normalize the track data. During the normalization process, position information, position information, speed scalar information and time information are normalized using the maximum and minimum values in the global range. The normalization standard formula is:

[0100] (5)

[0101] In the formula , , and respectively represent the in the at the position information, position information, speed scalar information and time information; , , and respectively represent the in the at the position information, position information, speed scalar information and time information. Statistically analyze the information of each dimension of all track segments in the scene. and respectively represent the minimum and maximum values of the target obtained statistically in the direction; and respectively represent the minimum and maximum values of the target obtained statistically in the direction; and respectively represent the minimum and maximum values of the speed scalar of the target obtained statistically. and respectively represent the minimum and maximum values of the time information of the target obtained statistically.

[0102] SS3, track prediction and association classification.

[0103] Traditional TSA algorithms use different track prediction strategies to extend old tracks to new tracks or predict both old and new tracks to the same time point to determine the similarity between two tracks. In non-dense environments, this method can achieve track association relatively accurately. However, in complex environments, due to the assumptions used in the track prediction process being difficult to fully adapt to the actual situation, the association accuracy drops significantly. This embodiment draws on the idea of the traditional algorithm to measure the similarity of track pairs through track prediction and constructs a framework for interrupted track association based on track prediction using deep learning theory to solve the problem of associating traditional interrupted tracks. The framework measures the similarity between the old and new tracks by predicting the old track to the new track and making a comparison, and determines whether the old and new tracks belong to the same target based on this. As Figure 2 shown, the framework mainly consists of three modules: an initial module, a track prediction module, and a classification module.

[0104] SS3.1, use the initial module to preprocess the track data of the old and new tracks.

[0105] The main function of the initial module is to smooth the track data and achieve feature dimension elevation. In interrupted track association, it is not only necessary to measure the similarity of the features of the old and new tracks, but also to analyze the differences between the two. After the initial module, by subtracting the variables of the extended old and new track data one by one, the differential features between the two track segments can be obtained. Subsequently, this difference is concatenated with the prediction result of the old track and the new track in the variable dimension, comprehensively considering the similarity and difference between the track segments, and the concatenated result is input into the classification module to finally obtain the probability of whether the track pair is associated.

[0106] Input track data usually contains observation noise and thus appears relatively "rough", making it difficult to effectively extract features. Using such "rough" data for track prediction will significantly reduce the prediction accuracy and have a direct impact on the final association performance. Traditional interrupted track association algorithms usually use an interactive multiple model algorithm based on Kalman filtering to track and filter the track data before track prediction to obtain a "smooth" track that is more in line with the true trajectory of the target. However, the processing flow of this method is relatively complex and highly dependent on the acquisition of prior information. In contrast, deep learning algorithms have strong feature extraction capabilities and can directly learn the internal patterns of target motion from track data without relying on prior information. Therefore, good results can be obtained as long as the track data is sufficiently "smooth". This embodiment uses mean filtering to "smooth" the input track data, and its calculation formula is:

[0107] (6)

[0108] where, is the mean filter window size. For track data with a longer time step, a larger filter window can be selected.

[0109] Given that the characteristic dimension of the input track data is small, in this embodiment, the first-order difference method is used to obtain new features, and its calculation formula is:

[0110] (7)

[0111] Subsequently, the first row of the filtered track data is removed, and it is concatenated with the difference result in the feature dimension. The track data processed by the above operations has a reduced time step length compared to the input track data However, its feature dimension doubles, thereby enhancing the feature extraction ability of the model. The shape change of the track data is as shown in Figure 3 where Figure 3 in represents the time step length of the original track data, and represents the number of features of the original track data.

[0112] SS3.2. Based on the preprocessed old track data, the track data of the predicted new track is obtained through a pre-trained track prediction model based on time series.

[0113] When performing track prediction, the entire track is regarded as a multivariate time series, and the old track is predicted and trained with the entire new track as the target, so that the prediction module can accurately map the old track to the new track. For associated new and old track pairs, the Euclidean distance between the prediction result of the old track and the new track on each variable should be as small as possible to ensure that the prediction result is closer to the new track; for unassociated track pairs, the Euclidean distance between the prediction result of the old track and the new track is relatively large.

[0114] In this embodiment, the S-Mamba model is selected as the track prediction module in the framework, and its overall structure is as shown in Figure 4 where Mamba, as the core component of the S-Mamba model, is constructed based on the Selective StateSpace Model (Selective SSM), aiming to efficiently process sequence dependencies while maintaining an approximately linear time complexity. This model performs excellently in capturing relevant information, especially having significant advantages in processing ultra-long time series data. Its model structure is as shown in Figure 5 as shown.

[0115] Among them, Selective SSM is based on the State Space Model (SSM). By parameterizing the input, the model can selectively process information. SSM is a mathematical method for time series modeling, mainly composed of a state equation and an output equation. Its specific form is as follows.

[0116] (8)

[0117] (9)

[0118] Among them, represents the latent state representation at a given time , represents the input at a given time , represents the output at a given time . A, B, C, and D represent the state transition matrix, input matrix, output matrix, and direct transmission matrix respectively.

[0119] Based on the track prediction model, the track data of the predicted new track is obtained through the following specific steps.

[0120] SS3.2.1, perform dimensional transformation on the input data through the first linear layer; the input data refers to the preprocessed old track data.

[0121] SS3.2.2, the output data of the first linear layer flows into the Mamba model and the reverse Mamba model respectively, and the sequence data is processed from the forward and reverse directions respectively.

[0122] SS3.2.3, merge the output data of the Mamba model and the reverse Mamba model through an addition operation.

[0123] SS3.2.4, normalize the merged data through the first normalization layer.

[0124] SS3.2.5, perform non-linear feature extraction on the output data of the first normalization layer through a feed-forward network layer.

[0125] SS3.2.6, perform an addition operation on the output data of the first normalization layer and the feed-forward network layer, and then input it into the second normalization layer for normalization processing.

[0126] SS3.2.7, input the output data of the second normalization layer into the second linear layer, and the second linear layer outputs the track data of the predicted new track.

[0127] SS3.3. Based on the preprocessed old track data, new track data, and predicted new track data, obtain the association classification result of the track pair through a pre-trained time series classification model.

[0128] By using the trained track prediction module to predict the expanded old track data, obtain its prediction result relative to the corresponding new track. For a new-old track pair that can be determined to be associated, the difference between its new and old track data should be significantly different from that of non-associated track pairs. At the same time, the similarity between the prediction result of the old track and the new track should be higher than that of non-associated track pairs.

[0129] In order to comprehensively express the difference and similarity between track pairs, in this embodiment, the difference between the expanded new and old track data, the prediction result, and the new track data are concatenated in the feature dimension to form a time series. The main function of the classification module is to use the time series classification model to extract features and classify this sequence. Through training, this module can classify and judge based on the difference and similarity between track pairs. For a new-old track pair to be determined, the more similar the prediction result of the old track is to the new track, the higher the probability that the track pair is determined to be associated. Combining the difference features of the track pair, the classification module can more accurately judge the association relationship between the new and old tracks.

[0130] During the target movement, it usually maintains a fixed movement state for a period of time, making the corresponding track data show a specific change trend during this period. These changes may be manifested as increasing or decreasing. The InceptionTime model is a deep learning network that combines the Inception module and residual connections, specifically designed for processing time series data. The H-InceptionTime model introduces three non-learnable one-dimensional general filters on the basis of the InceptionTime structure, which are used to detect the upward trend, downward trend, and peak in the time series, respectively, so as to effectively solve the time series classification problem. Based on this, this embodiment selects the H-InceptionTime model as the time series classification model in the classification module, and its overall model structure is as Figure 6 shown. Among them, the specific structure of the manually created filter part is as Figure 7 shown, and the specific structure of the Inception module is as Figure 8 shown.

[0131] Based on the time series classification model, the association classification of track pairs specifically includes the following steps.

[0132] SS3.3.1. Perform a subtraction operation on the preprocessed old track data and new track data.

[0133] SS3.3.3, perform a splicing operation on the subtraction operation result, the new track data, and the predicted new track data.

[0134] SS3.3.3, input the splicing result into a pre-trained time series classification model, and the time series classification model outputs the association classification result of the track pair.

[0135] SS3.3.3.1, make the splicing result flow into the increasing filter, the decreasing filter, the peak filter, and the first Inception module respectively.

[0136] SS3.3.3.2, perform a splicing operation on the output data of the increasing filter, the decreasing filter, the peak filter, and the first Inception module to obtain the integrated features.

[0137] SS3.3.3.3, the integrated features are passed through the second Inception module twice to extract multi-scale features.

[0138] SS3.3.3.4, perform the first residual connection on the input data and the extracted multi-scale features.

[0139] SS3.3.3.5, the result of the first residual connection is passed through the third Inception module three times to extract multi-scale features again.

[0140] SS3.3.3.6, perform the second residual connection on the result of the second residual connection and the multi-scale features extracted again.

[0141] SS3.3.3.7, reduce the feature dimension of the result of the second residual connection through the global average pooling layer.

[0142] SS3.3.3.8, pass the output data of the global average pooling layer through a linear layer to output the association classification result of the track pair.

[0143] In order to distinguish the track segments of different targets, it is necessary to design corresponding objective functions to optimize the network parameters. The higher the similarity between two track segments, the smaller the distance between them. The goal of the interrupted track association framework based on track prediction is: for two track segments determined to be associated, make the prediction result of the old track closer to the corresponding new track; for two track segments determined not to be associated, make the prediction result of the old track more different from the corresponding new track. In addition, it is also necessary to ensure that the model can accurately classify the spliced time series. Based on the above considerations, this embodiment designs two objective functions: one is for the track prediction model to optimize the distance between the prediction result of the old track and the new track; the other is for time series classification to optimize the classification result of the model.

[0144] 1) Distance optimization: Use contrastive loss to calculate the loss value between the prediction result of the old track and the new track during the training phase. Its definition is as follows:

[0145] (10)

[0146] where is a binary label. If the two track segments of the old track and the actual new track in the sample come from the same target, then ; otherwise . represents the interval distance, which is the shortest distance between two track segments belonging to different targets in the sample. represents the Euclidean distance between the predicted new track and the actual new track.

[0147] 2) Classification result optimization: Use the cross-entropy loss function to calculate the loss value between the model prediction probability and the true distribution. For the case of containing samples, assume the output of the model for each sample is , and the true label is 0 or 1. Then the corresponding cross-entropy loss function can be expressed as: Its definition is as follows:

[0148] (11)

[0149] The total loss function is described as follows:

[0150] (12)

[0151] SS4, reassign the association matrix according to the association classification result, and obtain the interrupted track association result based on the reassigned association matrix.

[0152] Update the values in the track association matrix according to the output result of the time series classification model. For the track pair , the output of the model is the probability that the old track and the new track do not belong to the same target and the probability that they belong to the same target. When and the current association flag , the model determines that this track pair is not associated and updates its value in the association matrix from 1 to 0, as shown in Equation (13)

[0153] (13)

[0154] Define the set as , which is used to store the serial numbers of the finally associated new and old tracks and is initialized to be empty.

[0155] (14)

[0156] After the time series classification model finishes its judgment, traverse the track pairs in the track association matrix that meet the conditions to find the old track with the largest value in each group of track pairs and the new track . Mark it as , and add this mark to the set . At this time, regard the old track and the new track as consecutive track segments of the same target. After selecting each pair of associated tracks, update the values of the corresponding elements in the association matrix , and set all the element values in the th row and the th column of the association matrix to 0. As shown in Eqs. (15)-(17).

[0157] (15)

[0158] (16)

[0159] (17)

[0160] Continue to perform the global maximum value traversal until the maximum value obtained from the traversal is less than or equal to 0 and stop. Finally, the content in the set is the result of track association.

[0161] The following explains the network training and parameter selection.

[0162] During the network training process, selecting an appropriate interval distance helps to better perform distance optimization. Based on this, an interval distance analysis experiment is carried out. Use the average recall rate and the optimal recall rate to represent the average performance and the optimal performance of the algorithm during the training process respectively. Assume that N rounds of training are carried out. The definition of

[0163] (18)

[0164] where K is the number of correctly associated new and old track pairs, is the number of correctly associated new and old track pairs by the algorithm after the th training. The definition of

[0165] (19)

[0166] Data preparation: Generate a dataset. Considering the movement of the target in a two-dimensional plane, select the CV, CA, CT, and CAT models as the simulation motion model set. The Markov state transition matrix between the target motion models is As shown in Equation (20).

[0167] (20)

[0168] Set the process noise during the target motion to follow a Gaussian distribution with a mean of 0 and a standard deviation of 0.1. The sensor position coordinates are , the angle measurement error , the range measurement error , the maximum speed of the target is , the minimum speed is , the maximum acceleration is , the minimum acceleration is , the maximum turning rate is , the minimum turning rate is , the total sampling duration is 100 s, the sampling interval is 1 s, and randomly generate the initial position of the target , randomly generate the initial heading , randomly generate the initial velocity . By setting the density control radius , and the density control center coordinates are , obtain a radar measurement area with a size of . Set the number of motion state switches of each target during this period to 5, and generate 5000 complete target tracks, as shown in Figure 9 .

[0169] Interval distance analysis: The interval distance is a key hyperparameter affecting the performance of TSAF-TP, which determines the minimum distinguishable degree between different target track segments. During the model training process, adjusting the value of will directly affect the discriminative ability of the model for similar track segments, thus affecting the classification accuracy and generalization performance. If 𝑚 is set too small, the model may not be able to effectively distinguish the track segments of different targets; if set too large, track segments that should be associated may be misjudged as uncorrelated. Therefore, it is necessary to evaluate different interval distances (0.4, 0.6, 0.8, 1.0, 1.2) to determine the optimal value.

[0170] Select the complete tracks of 500 targets for verification analysis, and set the interruption duration to 10 s. The experimental results of the average recall rate and the optimal recall rate of the model under different interval distances are shown in Table 1.

[0171] Table 1: Association performance under different interval distances

[0172]

[0173] As can be seen from Table 1, when the interval distance is 1.0, the performance of TSAF-TP is optimal. Then the interval distance of 1.0 will be used in subsequent experiments.

[0174] After that, algorithm verification and analysis were carried out. First, the network noise immunity was tested. It is very difficult to obtain an ideal noise-free dataset in a real environment. To explore the noise immunity performance of the method in this embodiment, the association effects in different noise scenarios were compared, and the noise levels in the real environment were simulated by setting different angle measurement errors and ranging errors.

[0175] 500 complete tracks of targets were selected for verification and analysis. The experimental results of the average recall rate and the optimal recall rate of the model under different noise levels with an interruption duration of 10 s are shown in Table 2.

[0176] Table 2: Association performance under different noises

[0177]

[0178] As can be seen from Table 2, the method in this embodiment shows good performance under different noises, which proves that the method in this embodiment has good noise immunity performance.

[0179] The comparative experiment is as follows: To describe the performance of the algorithm, the correct track association rate is selected as the performance evaluation index, and its calculation is shown in Equation 21

[0180] (21)

[0181] In the formula, represents the total number of track pairs that should be correctly associated in a scenario, represents the total number of correctly associated track pairs.

[0182] The correct association rate is selected as the comparative performance index. The state-dependent transition probability interactive multi-model interrupted track association (IMMSDP-TSA) algorithm, the track segment association algorithm based on the prior multi-hypothesis model (TSA-MHPI) algorithm, and the interrupted track association algorithm based on the metric neural network (TSADCNN) algorithm are selected as the comparison objects. The trained model is used to conduct performance comparison experiments with the above three algorithms, and the performance of each algorithm is compared under different numbers of targets and different interruption times.

[0183] Set the interruption duration to 10 s, the radar angle measurement error to , and the ranging error to , the influence of different numbers of targets on the performance of different TSA algorithms is shown in Table 3. The method of this embodiment always maintains a high correct association rate as the number of tracks increases.

[0184] Table 3: Correct association rates of each algorithm under different numbers of tracks

[0185]

[0186] Set the number of targets to 500, and the radar angle measurement error is , and the ranging error is , the influence of different interruption times on the performance of different TSA algorithms is shown in Table 4.

[0187] Table 4: Correct association rates of each algorithm under different interruption times

[0188]

[0189] As described in the above table, the increase in the interruption duration and the number of tracks will increase the association difficulty of each algorithm. Among them, the algorithm of this embodiment can maintain good performance in most cases.

[0190] In the above text, an embodiment of an interrupted track association method based on track prediction has been described in detail. Based on the interrupted track association method based on track prediction described in the above embodiment, the embodiment of the present invention also provides an interrupted track association system corresponding to this method.

[0191] Figure 10 FIG. is a schematic block diagram of the structure of an interrupted track association system based on track prediction provided by an embodiment of the present invention. In this embodiment, the interrupted track association system 1000 based on track prediction can be divided into multiple functional units according to the functions it performs. The unit referred to in the present invention means a series of computer program segments that can be executed by at least one processor and can complete fixed functions, and are stored in the memory.

[0192] The rough track association unit 1001 is used to perform rough association on all old tracks and new tracks in the scene based on time matching and speed matching, and assign values to the association matrix according to the rough association result.

[0193] The track data normalization processing unit 1002 is used to perform normalization processing on the track data of all track pairs after rough association.

[0194] The track data preprocessing unit 1003 is used to preprocess the track data of old tracks and new tracks, and the preprocessing includes mean filtering and variable dimension expansion operations.

[0195] The track data prediction unit 1004 is configured to obtain the track data for predicting a new track through a pre-trained time-series-based track prediction model based on the pre-processed old track data.

[0196] The association classification unit 1005 is configured to obtain the association classification result of the track pair through a pre-trained time-series classification model based on the pre-processed old track data, new track data, and predicted new track data.

[0197] The association result acquisition unit 1006 is configured to re-assign the association matrix according to the association classification result and obtain the interrupted track association result based on the re-assigned association matrix.

[0198] The interrupted track association system based on track prediction in this embodiment is used to implement the aforementioned interrupted track association method based on track prediction. Therefore, the specific implementation manners in this system can be seen in the embodiment part of the interrupted track association method based on track prediction in the foregoing text. Therefore, its specific implementation manners can be referred to the descriptions of the corresponding various part embodiments and will not be elaborated here.

[0199] In addition, since the interrupted track association system based on track prediction in this embodiment is used to implement the aforementioned interrupted track association method based on track prediction, its functions correspond to those of the above method and will not be elaborated here.

[0200] Figure 11 The following is a schematic structural diagram of a terminal 1100 provided by an embodiment of the present invention, including: a processor 1110, a memory 1120, and a communication unit 1130. When the processor 1110 implements the interrupted track association program based on track prediction stored in the memory 320, the following steps are implemented:

[0201] Perform rough association on all old tracks and new tracks in the scene based on time matching and speed matching, and assign values to the association matrix according to the rough association result;

[0202] Perform normalization processing on the track data of all track pairs after rough association;

[0203] Perform pre-processing on the track data of old tracks and new tracks, and the pre-processing includes mean filtering and variable dimension expansion operations;

[0204] Based on the pre-processed old track data, obtain the track data for predicting a new track through a pre-trained time-series-based track prediction model;

[0205] Based on the pre-processed old track data, new track data, and predicted new track data, obtain the association classification result of the track pair through a pre-trained time-series classification model;

[0206] Reassign the association matrix according to the associated classification result, and obtain the interrupted track association result based on the reassigned association matrix.

[0207] The present invention also provides a computer storage medium, and the storage medium herein may be a magnetic disk, an optical disk, a read-only memory (ROM for short), a random access memory (RAM for short), etc.

[0208] The computer storage medium stores an interrupted track association program based on track prediction. When the interrupted track association program based on track prediction is executed by a processor, the following steps are implemented:

[0209] Perform rough association on all old tracks and new tracks in the scene based on time matching and speed matching, and assign values to the association matrix according to the rough association result;

[0210] Normalize the track data of all track pairs after rough association;

[0211] Preprocess the track data of old tracks and new tracks. The preprocessing includes mean filtering and variable dimension expansion operations;

[0212] Based on the preprocessed old track data, obtain the track data of the predicted new track through a pre-trained track prediction model based on time series;

[0213] Based on the preprocessed old track data, new track data, and predicted new track data, obtain the associated classification result of the track pair through a pre-trained time series classification model;

[0214] Reassign the association matrix according to the associated classification result, and obtain the interrupted track association result based on the reassigned association matrix.

[0215] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An interrupted track association method based on track prediction, characterized in that, It includes the following steps: Roughly associate all the old tracks and new tracks in the scene based on time matching and speed matching, and assign values to the association matrix according to the rough association results; Normalize the track data of all track pairs after rough association; Preprocess the track data of the old tracks and new tracks, and the preprocessing includes mean filtering and variable dimension expansion operations; Based on the preprocessed old track data, obtain the track data of the predicted new track through a pre-trained track prediction model based on time series; Based on the preprocessed old track data, new track data, and predicted new track data, obtain the association classification result of the track pair through a pre-trained time series classification model. Specifically, it includes: subtracting the preprocessed old track data and new track data, concatenating the subtraction operation result, new track data, and predicted new track data, inputting the concatenation result into the pre-trained time series classification model, and outputting the association classification result of the track pair by the time series classification model; Reassign values to the association matrix according to the association classification result, and obtain the interrupted track association result based on the reassigned association matrix; Among them, inputting the concatenation result into the pre-trained time series classification model and outputting the association classification result of the track pair by the time series classification model specifically includes: Making the concatenation result flow into the increasing filter, decreasing filter, peak filter, and the first Inception module respectively; Performing a concatenation operation on the output data of the increasing filter, decreasing filter, peak filter, and the first Inception module to obtain the integrated feature; The integrated feature passes through the second Inception module twice to extract multi-scale features; Performing the first residual connection on the input data and the extracted multi-scale features; The result of the first residual connection passes through the third Inception module three times to extract multi-scale features again; Performing the second residual connection on the result of the second residual connection and the multi-scale features extracted again; Reducing the feature dimension of the result of the second residual connection through the global average pooling layer; Passing the output data of the global average pooling layer through a linear layer to output the association classification result of the track pair.

2. The method for associating interrupted tracks based on track prediction according to claim 1, wherein Based on the preprocessed old track data, obtain the track data of the predicted new track through a pre-trained track prediction model based on time series. Specifically, it includes: Performing a dimension transformation on the input data through the first linear layer; the input data refers to the preprocessed old track data; The output data of the first linear layer flows into the Mamba model and the reverse Mamba model respectively, and processes the sequence data from the positive and reverse directions respectively; Combining the output data of the Mamba model and the reverse Mamba model through an addition operation; Normalizing the combined data through the first normalization layer; Performing non-linear feature extraction on the output data of the first normalization layer through a feed-forward network layer; Performing an addition operation on the output data of the first normalization layer and the feed-forward network layer and then inputting it into the second normalization layer for normalization; Inputting the output data of the second normalization layer into the second linear layer, and outputting the track data of the predicted new track by the second linear layer.

3. The method for associating interrupted tracks based on track prediction according to claim 1 or 2, characterized in that, The objective function of the track prediction model in the training phase is to use contrastive loss to calculate the loss value between the predicted new track and the actual new track, which is defined as: Among them, is a binary label. If the two track segments of the old track and the actual new track in the sample come from the same target, then , otherwise ; represents the interval distance, which is the shortest distance between two track segments belonging to different targets in the sample; represents the Euclidean distance between the predicted new track and the actual new track.

4. The method for associating interrupted tracks based on track prediction according to claim 1, wherein The objective function of the time series classification model in the training phase is to use the cross-entropy loss function to calculate the loss value between the predicted probability and the true distribution. The output of the time series classification model for each sample is , is the probability that the old track and the new track do not belong to the same target, is the probability that the old track and the new track belong to the same target. The true label is 0 or 1, and the corresponding cross-entropy loss function is expressed as: wherein is the number of samples; The total loss function is expressed as: .

5. The method for associating interrupted tracks based on track prediction according to claim 1, wherein Reassign the association matrix according to the association classification result, and obtain the interrupted track association result based on the reassigned association matrix, specifically including: The associated classification results include the old track and the new track of not belonging to the same target and the probability of belonging to the same target ; The element in the -th row and -th column of the association matrix is the association mark between the -th old track and the -th new track ; indicates that the two are not associated, indicates that the two are associated; When and the current associated flag is, update the associated flag to 0; Define a set and initialize it to be empty, which is used to store the serial numbers of the new and old tracks finally associated; Traverse the track pairs in the association matrix that satisfy , and find the old track serial number that is in each track pair and has the largest value, as well as the new track , and mark them as ; Add the tag to the set , and set all elements in the th row and the th column of the association matrix to 0, then traverse the track pairs in the association matrix that satisfy again; finally, the content in the set is the interrupted track association result.

6. An interrupted track association system based on track prediction, characterized in that, Including: A track rough association unit, which is used to roughly associate all old tracks and new tracks in the scene based on time matching and speed matching, and assign values to the association matrix according to the rough association result; A track data normalization processing unit, which is used to normalize the track data of all track pairs after rough association; A track data preprocessing unit, which is used to preprocess the track data of old tracks and new tracks, and the preprocessing includes mean filtering and variable dimension expansion operations; A track data prediction unit, which is used to obtain the track data of the predicted new track based on the preprocessed old track data through a pre-trained track prediction model based on time series; An association classification unit, which is used to obtain the association classification result of the track pair through a pre-trained time series classification model based on the preprocessed old track data, new track data, and predicted new track data. Specifically, it includes: subtracting the preprocessed old track data from the new track data, concatenating the subtraction result, the new track data, and the predicted new track data, and inputting the concatenated result into the pre-trained time series classification model, and the time series classification model outputs the association classification result of the track pair; An association result acquisition unit, which is used to reassign the association matrix according to the association classification result and obtain the interrupted track association result; Among them, inputting the concatenated result into the pre-trained time series classification model, and the time series classification model outputs the association classification result of the track pair, specifically including: Making the concatenated result flow into an increasing filter, a decreasing filter, a peak filter, and the first Inception module respectively; Performing a concatenation operation on the output data of the increasing filter, the decreasing filter, the peak filter, and the first Inception module to obtain the integrated feature; The integrated feature is extracted by the second Inception module twice to obtain multi-scale features; Performing the first residual connection on the input data and the extracted multi-scale features; The result of the first residual connection is extracted by the third Inception module three times to obtain multi-scale features again; Performing the second residual connection on the result of the second residual connection and the multi-scale features extracted again; Reducing the feature dimension of the result of the second residual connection through a global average pooling layer; Outputting the association classification result of the track pair through a linear layer for the output data of the global average pooling layer.

7. A terminal, characterized in that, Including: A memory, which is used to store the interrupted track association program based on track prediction; A processor, which is used to implement the steps of the interrupted track association method based on track prediction as described in any one of claims 1 to 5 when executing the interrupted track association program based on track prediction.

8. A computer-readable storage medium, characterized in that, The readable storage medium stores an interrupted track association program based on track prediction. When the interrupted track association program based on track prediction is executed by a processor, it implements the steps of the interrupted track association method based on track prediction according to any one of claims 1 to 5.

Images