Vehicle trajectory tracking method, system and device and storage medium

Abstract

The invention provides a vehicle trajectory tracking method, system and device, and a storage medium. The method comprises the following steps: preliminarily merging trajectories corresponding to a plurality of IDs endowed to the same vehicle to obtain an initial trajectory set of all vehicles; grouping the initial track set according to road sections, and performing track fitting on the target vehicle in each road section; determining an abnormal track based on a fitting result; matching the abnormal trajectory with candidate trajectories except the initial trajectory set in the same road section; and in response to successful matching, fusing the successfully matched candidate trajectory with the abnormal trajectory to form the trajectory of the road section, and combining the trajectory of each road section to form a final trajectory. According to the method, the problem of ID hopping caused by passing through different sensing areas, frame loss, blind areas and the like is solved through preliminary merging and road section track fitting.

Description

Technical Field

[0001] The present invention relates to the field of vehicle-road cooperation, and more specifically, to a method, system, device and storage medium for vehicle trajectory tracking. Background Art

[0002] Traditional multi-object tracking algorithms calculate the IOU (Intersection over Union) of objects through the Hungarian matching algorithm by comparing objects in the front and rear frames of the sensed image to achieve object re-ID (marking the same object with the same ID). This method is suitable for short-distance object tracking. To perform long-distance tracking, two problems need to be faced: when a vehicle travels from one sensing area to another, ID jumps will occur, resulting in discontinuous driving trajectories; blind spots and signal frame loss will also cause the recognition service to not receive continuous vehicle position information, resulting in ID (identification) jumps, which easily cause the trajectory in the area corresponding to the ID jump to disappear, that is, the trajectory information of this area cannot be obtained through fitting, and this area is also called the "area where the trajectory cannot be fitted". The existence of the "area where the trajectory cannot be fitted" leads to the discontinuity of the vehicle driving trajectory, forming abnormal trajectories, and the continuous and complete trajectory during the vehicle driving cannot be obtained. Summary of the Invention

[0003] In view of this, the purpose of the embodiments of the present invention is to provide a method, system, electronic device and computer-readable storage medium for vehicle trajectory tracking. The present invention is no longer limited to tracking vehicles through the front and rear frames of images. By initially merging and fitting the trajectories of different sections, the problem of ID jumps caused by crossing different sensing areas, frame loss, blind spots, etc. is solved, and a continuous and complete trajectory is obtained.

[0004] Based on the above purpose, on the one hand, an embodiment of the present invention provides a method for vehicle trajectory tracking, including the following steps: initially merging the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain an initial trajectory set of all vehicles; grouping the initial trajectory set according to sections, and fitting the trajectories of target vehicles in each section; determining abnormal trajectories based on the fitting results, and matching the abnormal trajectories with candidate trajectories outside the initial trajectory set in the same section; and in response to a successful match, fusing the successfully matched candidate trajectories with the abnormal trajectories to form the trajectory of the section, and combining the trajectories of each section to form the final trajectory.

[0005] In some embodiments, the step of initially merging the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain the initial trajectory set of all vehicles includes: obtaining the bounding box coordinates of the vehicle corresponding to each ID to form a bounding box, determining whether the vehicles corresponding to different IDs are the same vehicle by calculating the ratio of the overlapping area of the bounding boxes of the vehicles corresponding to two IDs, and merging the trajectories of the same vehicle, and taking the trajectories of all the merged vehicles as the initial trajectory set of all vehicles.

[0006] In some embodiments, the step of fitting the trajectory of the target vehicle in each section includes: obtaining the license plate data of the target vehicle, and matching the trajectory corresponding to the target vehicle in the initial trajectory set according to the license plate data.

[0007] In some embodiments, the step of fitting the trajectory of the target vehicle in each section includes: obtaining the three-dimensional position coordinates of the target vehicle, and projecting the three-dimensional position coordinates into a plane coordinate system through spatial transformation to match the license plate data and the trajectory corresponding to the target vehicle.

[0008] In some embodiments, the step of matching the abnormal trajectory with candidate trajectories other than the initial trajectory set in the same section includes: dividing the time when passing through the section into multiple time windows, and matching the abnormal trajectory and the candidate trajectories in the same time window in the section according to the characteristics of the abnormal trajectory. Among them, the characteristics of the abnormal trajectory include one or more of license plate similarity, vehicle color, vehicle size, lane information, vehicle movement speed and direction.

[0009] In some embodiments, the step of determining whether the abnormal trajectory and the candidate trajectories in the same time window match according to the multiple feature columns includes: training a machine learning algorithm model through a historical data set to determine the mapping relationship between the multiple features and the matching result, extracting the features of the abnormal trajectory according to the trained machine learning algorithm model, and determining whether the abnormal trajectory and the candidate trajectories in the same time window match according to the similarity of the features of different abnormal trajectories.

[0010] In some embodiments, the step of determining the abnormal trajectory includes: taking the trajectories that disappear midway and the trajectories that newly appear midway in the section as the abnormal trajectories in the section.

[0011] On the other hand, an embodiment of the present invention provides a vehicle trajectory tracking system, including: a merging module configured to preliminarily merge the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain an initial trajectory set of all vehicles; a grouping module configured to group the initial trajectory set by road segments and perform trajectory fitting on the target vehicles in each road segment; a matching module configured to determine abnormal trajectories based on the fitting results, and match the abnormal trajectories with candidate trajectories outside the initial trajectory set in the same road segment; and an execution module configured to, in response to a successful match, fuse the successfully matched candidate trajectories with the abnormal trajectories to form the trajectory of the road segment, and combine the trajectories of each road segment to form a final trajectory.

[0012] In yet another aspect, an embodiment of the present invention further provides an electronic device, including: at least one processor; and a memory storing computer instructions executable on the processor, and when the instructions are executed by the processor, the steps of the above method are implemented.

[0013] In still another aspect, an embodiment of the present invention further provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above method.

[0014] The present invention has the following beneficial technical effects: It is no longer limited to tracking vehicles through consecutive frames of images. On the one hand, by preliminarily merging the trajectories corresponding to multiple IDs assigned to the same vehicle, the ID jump across different sensing areas is solved; by performing trajectory fitting for each road segment, the ID jump caused by reasons such as frame loss and blind spots is solved, and finally a continuous trajectory during vehicle driving can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other embodiments based on these drawings without creative efforts.

[0016] Figure 1 It is a schematic diagram of an embodiment of the method for vehicle trajectory tracking provided by the present invention;

[0017] Figure 2 It is a schematic diagram of an embodiment of the vehicle trajectory tracking system provided by the present invention;

[0018] Figure 3 It is a schematic diagram of the hardware structure of an embodiment of the electronic device for vehicle trajectory tracking provided by the present invention;

[0019] Figure 4 Schematic diagram of an embodiment of a computer storage medium for vehicle trajectory tracking provided by the present invention. Detailed implementation manners

[0020] To make the objectives, technical solutions and advantages of the present invention more clear and understandable, the following further describes the embodiments of the present invention in detail with reference to specific embodiments and the accompanying drawings.

[0021] It should be noted that in all the descriptions of the embodiments of the present invention, the expressions using "first" and "second" are only used to distinguish two entities or parameters with the same name but different identities. It can be seen that "first" and "second" are only for the convenience of expression and should not be construed as a limitation on the embodiments of the present invention. This will not be elaborated one by one in the subsequent embodiments.

[0022] In the first aspect of the embodiments of the present invention, an embodiment of a method for vehicle trajectory tracking is proposed. Figure 1 Shown is a schematic diagram of an embodiment of a method for vehicle trajectory tracking provided by the present invention.

[0023] As Figure 1 shown, the embodiments of the present invention include the following steps:

[0024] S1. Initially merge the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain an initial trajectory set of all vehicles;

[0025] S2. Group the initial trajectory set by road segments, and perform trajectory fitting on the target vehicles in each road segment;

[0026] S3. Determine abnormal trajectories based on the fitting results;

[0027] S4. Match the abnormal trajectories with candidate trajectories outside the initial trajectory set in the same road segment;

[0028] S5. In response to a successful match, fuse the successfully matched candidate trajectories with the abnormal trajectories to form the trajectory of the road segment; and

[0029] S6. Combine the trajectories of each road segment to form the final trajectory.

[0030] In the field of vehicle-road collaboration, information such as the position, speed, and shape of traffic participants in the road can be obtained through roadside lidar and cameras. By tracking the trajectories of vehicles, the driving behaviors of each vehicle can be analyzed, and road events can be identified. It provides rich scenarios for autonomous driving simulation. By analyzing the driving behaviors of vehicles, the driving behavior characteristics and portraits of vehicles can be obtained, but these all rely on the long-distance tracking of vehicle trajectories.

[0031] During the process of image acquisition by the camera, the same object may be recognized as multiple IDs. For example, when passing through a blind spot of perception, experiencing serious sensor frame loss, or passing through the overlapping area between two large groups, it may be different in different large groups. The continuous spatio-temporal data of one ID is a trajectory, and one driving of an object will be segmented into multiple trajectories. The goal of the embodiments of the present invention is to restore these multiple trajectories into one.

[0032] Preliminarily merge the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain the initial trajectory set of all vehicles.

[0033] In some embodiments, the step of preliminarily merging the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain the initial trajectory set of all vehicles includes: obtaining the bounding box coordinates of the vehicle corresponding to each ID to form a bounding box, determining whether the vehicles corresponding to different IDs are the same vehicle by calculating the ratio of the overlapping area of the bounding boxes of the vehicles corresponding to two IDs, and merging the trajectories of the same vehicle. The trajectories of all vehicles after merging are used as the initial trajectory set of all vehicles. Simply put, when the ratio of the overlapping area of the bounding boxes of two vehicles corresponding to two IDs that appear separately in adjacent sensing areas is greater than the threshold, it can be considered that the two vehicles corresponding to the two IDs are the same vehicle. Accordingly, the trajectories of the same object in the adjacent sensing areas can be preliminarily merged. The trajectory segments that fail to be incorporated into the initial trajectory set can be used as candidate trajectories in subsequent steps.

[0034] Group the initial trajectory set by road section, and perform trajectory fitting on the target vehicles in each road section. Integrate the vehicle coordinates with the high-precision map to obtain the road section and lane information of the vehicle. Group the trajectories in the initial trajectory set sub1 by road section and then perform trajectory fitting for each road section.

[0035] In some embodiments, the step of performing trajectory fitting on the target vehicles in each road section includes: obtaining the license plate data of the target vehicle, and matching the trajectory corresponding to the target vehicle in the initial trajectory set according to the license plate data. The license plate data can be obtained through a camera.

[0036] In some embodiments, the step of performing trajectory fitting on the target vehicles in each road section includes: obtaining the three-dimensional position coordinates of the target vehicle, and projecting the three-dimensional position coordinates onto a plane coordinate system through a spatial transformation to match the license plate data and the trajectory corresponding to the target vehicle. The license plate data is from the camera at the intersection, and the vehicle position and speed coordinate information is from the lidar. Multisensor data fusion is required. Project the vehicle 3D coordinate system coordinates onto the camera 2D plane coordinate system through a spatial transformation to achieve the matching of the vehicle trajectory and the license plate.

[0037] Determine abnormal trajectories based on the fitting results, and match the abnormal trajectories in the same time window in the section with candidate trajectories other than the initial trajectory set in the same section according to the characteristics of the abnormal trajectories. In the images captured by the camera, it cannot be guaranteed that the license plate data can be clearly captured in each frame. Therefore, there are some images in which no valid license plate data can be read, and the trajectories corresponding to these images are difficult to fit. If the trajectories of the same vehicle in a section cannot form a continuous whole, it means that the trajectories of the vehicle in this section cannot be fitted.

[0038] In some embodiments, the abnormal trajectory is specifically: a trajectory in which the driving trajectory of the vehicle is interrupted due to the existence of an area where the trajectory cannot be fitted. The step of determining the abnormal trajectory includes: regarding the trajectories that disappear midway and the trajectories that newly appear midway in the section as the abnormal trajectories in the section. In a section, divide it into time windows (for example, one window every 10 seconds), find the trajectories sub2 that disappear midway (excluding the trajectories that drive out of the section interval) and the trajectories sub3 that newly appear midway (excluding the trajectories that drive into the section interval), then sub2 and sub3 are both candidate abnormal trajectories (because in a closed section, vehicles cannot disappear out of thin air or appear out of thin air).

[0039] For example, the two trajectory data are as follows:

[0040]

[0041]

[0042] Trajectory id1 and trajectory id2 are in the same section. Trajectory id1 ended at 9:32:16, while trajectory id2 started at 9:35:19, with an interval of 3 seconds in between. Moreover, combining the map data, the end point of trajectory id1 is not at the section boundary and the start point of trajectory id2 is not at the section boundary either. Then trajectory id1 and trajectory id2 belong to the candidates for fusion.

[0043] In some embodiments, the step of matching the abnormal trajectory with candidate trajectories other than the initial trajectory set in the same section includes: dividing the time when passing through the section into multiple time windows, and matching the abnormal trajectory and the candidate trajectory in the same time window in the section according to the characteristics of the abnormal trajectory. Among them, multiple feature columns are formed according to the license plate similarity, vehicle color, vehicle size, lane information, vehicle movement speed and direction included in the abnormal trajectory, and it is determined whether the abnormal trajectory and the candidate trajectory in the same time window match according to the multiple feature columns.

[0044] In some embodiments, the step of determining whether the abnormal trajectory and the candidate trajectory in the same time window match according to the multiple feature columns includes: training a machine learning algorithm model through a historical data set to determine the mapping relationship between the multiple features and the matching result, extracting the features of the abnormal trajectory according to the trained machine learning algorithm model, and determining whether the abnormal trajectory and the candidate trajectory in the same time window match according to the similarity of the features of different abnormal trajectories.

[0045] Match the trajectories in sub2 and sub3. Based on license plate similarity, vehicle color, vehicle size, lane information, combined with vehicle movement speed and direction, perform trajectory prediction to form multiple feature columns. If the two trajectories can be successfully matched, mark it as 1, otherwise mark it as 0. Then the trajectory matching problem becomes a binary classification problem based on a series of features. Train a machine learning algorithm model through a manually labeled data set to find the mapping relationship between these features and the matching result. Finally, through algorithm prediction, the two trajectories are matched to obtain trajectory sub4.

[0046] For example, when comparing trajectory id1 and trajectory id2 based on the data from the previous step, the following features are obtained: license plate similarity 0.8; vehicle size matching degree 0.99; whether the lane numbers are the same 1; whether the colors are the same 1; direction included angle 0.01; position prediction deviation 0 (the distance deviation between the position calculated according to the end velocity direction and time and the start position of the next trajectory). Based on the above feature model, it is predicted that trajectory id1 and trajectory id2 belong to the same vehicle.

[0047] In response to a successful match, fuse the successfully matched candidate trajectory with the abnormal trajectory to form the trajectory of the section, and combine the trajectories of each section to form the final trajectory. Perform the following fusion on the trajectory set sub4 of each target vehicle. Merge multiple trajectories into one according to the uniqueness of the license plate to achieve reID over a long spatio-temporal distance.

[0048] It should be particularly noted that each step in each embodiment of the above vehicle trajectory tracking method can be mutually crossed, replaced, added, or deleted. Therefore, these reasonable permutation and combination transformations for the vehicle trajectory tracking method should also fall within the protection scope of the present invention, and the protection scope of the present invention should not be limited to the embodiments.

[0049] Based on the above objectives, the second aspect of the embodiments of the present invention proposes a vehicle trajectory tracking system. As Figure 2As shown in the figure, the system 200 includes the following modules: a merging module configured to preliminarily merge the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain an initial trajectory set of all vehicles; a grouping module configured to group the initial trajectory set by road segments and perform trajectory fitting on the target vehicles in each road segment; a matching module configured to determine abnormal trajectories based on the fitting results, and match the abnormal trajectories with candidate trajectories outside the initial trajectory set in the same road segment; and an execution module configured to, in response to a successful match, fuse the successfully matched candidate trajectories with the abnormal trajectories to form the trajectory of the road segment, and combine the trajectories of each road segment to form the final trajectory.

[0050] In some embodiments, the merging module is further configured to: obtain the bounding box coordinates of the vehicle corresponding to each ID to form a bounding box, determine whether the vehicles corresponding to different IDs are the same vehicle by calculating the ratio of the overlapping area of the bounding boxes of the vehicles corresponding to two IDs, and merge the trajectories of the same vehicle, and use the merged trajectories of all vehicles as the initial trajectory set of all vehicles.

[0051] In some embodiments, the grouping module is further configured to: obtain the license plate data of the target vehicle and match the trajectory corresponding to the target vehicle in the initial trajectory set according to the license plate data.

[0052] In some embodiments, the grouping module is further configured to: obtain the three-dimensional position coordinates of the target vehicle, project the three-dimensional position coordinates into a plane coordinate system through a spatial transformation, so as to match the license plate data with the trajectory corresponding to the target vehicle.

[0053] In some embodiments, the matching module is further configured to: form multiple feature columns according to the license plate similarity, vehicle color, vehicle size, lane information, vehicle movement speed and direction included in the abnormal trajectory, and determine whether the abnormal trajectory and the candidate trajectory in the same time window match according to the multiple feature columns.

[0054] In some embodiments, the matching module is further configured to: train a machine learning algorithm model through a historical data set to determine the mapping relationship between the multiple features and the matching result, extract the features of the abnormal trajectory according to the trained machine learning algorithm model, and determine whether the abnormal trajectory and the candidate trajectory in the same time window match according to the similarity of the features of different abnormal trajectories.

[0055] In some embodiments, the matching module is further configured to: regard the trajectories that disappear midway and the trajectories that newly appear midway in the road segment as the abnormal trajectories in the road segment.

[0056] For the above purposes, in the third aspect of the embodiments of the present invention, an electronic device is proposed, including: at least one processor; and a memory storing computer instructions that can run on the processor, and the instructions are executed by the processor to implement the following steps: S1. Initially merge the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain the initial trajectory set of all vehicles; S2. Group the initial trajectory set by road segments, and perform trajectory fitting on the target vehicles in each road segment; S3. Determine the abnormal trajectories in the road segment based on the fitting results; S4. Match the abnormal trajectories with candidate trajectories outside the initial trajectory set in the same road segment; S5. In response to a successful match, fuse the successfully matched candidate trajectories with the abnormal trajectories to form the trajectory of the road segment; and S6. Combine the trajectories of each road segment to form the final trajectory.

[0057] In some embodiments, the step of initially merging the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain the initial trajectory set of all vehicles includes: obtaining the bounding box coordinates of the vehicle corresponding to each ID to form a bounding box, determining whether the vehicles corresponding to different IDs are the same vehicle by calculating the ratio of the overlapping area of the bounding boxes of the vehicles corresponding to two IDs, and merging the trajectories of the same vehicle, and taking the trajectories of all merged vehicles as the initial trajectory set of all vehicles.

[0058] In some embodiments, the step of performing trajectory fitting on the target vehicles in each road segment includes: obtaining the license plate data of the target vehicle, and matching the trajectory corresponding to the target vehicle in the initial trajectory set according to the license plate data. Among them, the license plate data of the vehicle can be obtained through a camera.

[0059] In some embodiments, the step of performing trajectory fitting on the target vehicles in each road segment includes: obtaining the three-dimensional position coordinates of the target vehicle, and projecting the three-dimensional position coordinates into a plane coordinate system through a spatial transformation to match the license plate data and the trajectory corresponding to the target vehicle. Among them, the three-dimensional position coordinates of the vehicle can be obtained through a lidar.

[0060] In some embodiments, the step of matching the abnormal trajectories with candidate trajectories outside the initial trajectory set in the same road segment includes: dividing the time when passing through the road segment into multiple time windows, and matching the abnormal trajectories and the candidate trajectories in the same time window in the road segment according to the characteristics of the abnormal trajectories; forming multiple feature columns according to the license plate similarity, vehicle color, vehicle size, lane information, vehicle movement speed and direction included in the abnormal trajectories, and determining whether the abnormal trajectories and the candidate trajectories in the same time window match according to the multiple feature columns.

[0061] In some embodiments, the step of determining whether the abnormal trajectory and the candidate trajectory in the same time window match according to the multiple feature columns includes: training a machine learning algorithm model through a historical data set to determine the mapping relationship between the multiple features and the matching result, extracting the features of the abnormal trajectory according to the trained machine learning algorithm model, and determining whether the abnormal trajectory and the candidate trajectory in the same time window match according to the similarity of the features of different abnormal trajectories.

[0062] In some embodiments, the step of determining the abnormal trajectory in the road segment includes: regarding the trajectory that disappears midway and the trajectory that newly appears midway in the road segment as the abnormal trajectory in the road segment.

[0063] As Figure 3 shown, it is a schematic diagram of the hardware structure of an embodiment of the above-mentioned electronic device for vehicle trajectory tracking provided by the present invention.

[0064] Taking the device as Figure 3 shown as an example, in this device, there is a processor 301 and a memory 302.

[0065] The processor 301 and the memory 302 can be connected through a bus or other means, Figure 3 and taking the connection through the bus as an example.

[0066] The memory 302, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the method for vehicle trajectory tracking in the embodiments of the present application. The processor 301 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions, and modules stored in the memory 302, that is, implements the method for vehicle trajectory tracking.

[0067] The memory 302 can include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the method for vehicle trajectory tracking, etc. In addition, the memory 302 can include a high-speed random access memory, and can also include non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some embodiments, the memory 302 can optionally include a memory remotely set relative to the processor 301, and these remote memories can be connected to the local module through a network. Examples of the above-mentioned network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0068] One or more computer instructions 303 corresponding to the method of vehicle trajectory tracking are stored in the memory 302, and when executed by the processor 301, they execute the method of vehicle trajectory tracking in any of the above method embodiments.

[0069] Any embodiment of the electronic device that executes the above method of vehicle trajectory tracking can achieve the same or similar effects as any of the foregoing method embodiments corresponding thereto.

[0070] The present invention also provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program that, when executed by a processor, executes the method of vehicle trajectory tracking.

[0071] As Figure 4 shown, it is a schematic diagram of an embodiment of the above computer storage medium for vehicle trajectory tracking provided by the present invention. Taking the computer storage medium as Figure 4 shown, the computer-readable storage medium 401 stores a computer program 402 that, when executed by a processor, executes the above method.

[0072] Finally, it should be noted that those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing relevant hardware through a computer program. The program of the method of vehicle trajectory tracking can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above method embodiments. Among them, the storage medium of the program can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc. The above embodiments of the computer program can achieve the same or similar effects as any of the foregoing method embodiments corresponding thereto.

[0073] The above are exemplary embodiments disclosed by the present invention. However, it should be noted that various changes and modifications can be made without departing from the scope of the embodiments disclosed by the present invention as defined by the claims. The functions, steps, and / or actions of the method claims according to the disclosed embodiments herein do not need to be executed in any specific order. In addition, although the elements disclosed in the embodiments of the present invention can be described or claimed in an individual form, they can also be understood as plural unless explicitly limited to the singular.

[0074] It should be understood that, as used herein, unless the context clearly supports exceptions, the singular form "a" is also intended to include the plural form. It should also be understood that the "and / or" used herein refers to any and all possible combinations of one or more of the related listed items.

[0075] The above serial numbers of the disclosed embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0076] Those of ordinary skill in the art can understand that all or part of the steps to implement the above embodiments can be completed by hardware, or can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable storage medium, and the above-mentioned storage medium can be a read-only memory, a magnetic disk, an optical disc, or the like.

[0077] Those of ordinary skill in the art should understand that the discussion of any above embodiment is only exemplary, and is not intended to imply that the scope (including the claims) disclosed by the embodiments of the present invention is limited to these examples; under the idea of the embodiments of the present invention, the technical features in the above embodiments or different embodiments can also be combined, and there are many other variations in different aspects of the above embodiments of the present invention, which are not provided in detail for the sake of brevity. Therefore, any omission, modification, equivalent replacement, improvement, etc. made within the spirit and principle of the embodiments of the present invention shall be included in the protection scope of the embodiments of the present invention.

Claims

1. A method for vehicle trajectory tracking, characterized in that, It includes the following steps: Preliminarily merge the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain the initial trajectory set of all vehicles; Group the initial trajectory set by road segments, and perform trajectory fitting on the target vehicles in each road segment; Determine abnormal trajectories based on the fitting results; Match the abnormal trajectories with candidate trajectories outside the initial trajectory set in the same road segment; In response to a successful match, fuse the successfully matched candidate trajectories with the abnormal trajectories to form the trajectory of the road segment; And Combine the trajectories of each road segment to form the final trajectory.

2. The method for vehicle trajectory tracking according to claim 1, characterized in that, The step of preliminarily merging the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain the initial trajectory set of all vehicles includes: Obtain the bounding box coordinates of the vehicle corresponding to each ID to form a bounding box, determine whether the vehicles corresponding to different IDs are the same vehicle by calculating the ratio of the overlapping area of the bounding boxes of the vehicles corresponding to two IDs, and merge the trajectories of the same vehicle, and use the trajectories of all merged vehicles as the initial trajectory set of all vehicles.

3. The method for vehicle trajectory tracking according to claim 1, characterized in that, The step of performing trajectory fitting on the target vehicles in each road segment includes: Obtain the license plate data of the target vehicle, and match the trajectory corresponding to the target vehicle in the initial trajectory set according to the license plate data.

4. The method for vehicle trajectory tracking according to claim 3, characterized in that, The step of performing trajectory fitting on the target vehicles in each road segment includes: Obtain the three-dimensional position coordinates of the target vehicle, and project the three-dimensional position coordinates into a plane coordinate system through spatial transformation to match the license plate data and the trajectory corresponding to the target vehicle.

5. The method for vehicle trajectory tracking according to claim 1, characterized in that, The step of matching the abnormal trajectories with candidate trajectories outside the initial trajectory set in the same road segment includes: Divide the time when passing through the road segment into multiple time windows, and match the abnormal trajectories and the candidate trajectories in the same time window in the road segment according to the characteristics of the abnormal trajectories; Among them, the characteristics of the abnormal trajectories include one or more of license plate similarity, vehicle color, vehicle size, lane information, vehicle movement speed and direction.

6. The method for vehicle trajectory tracking according to claim 5, characterized in that, The step of matching the abnormal trajectories with candidate trajectories outside the initial trajectory set in the same road segment includes: Train a machine learning algorithm model through a historical data set to determine the mapping relationship between the multiple characteristics and the matching results, extract the characteristics of the abnormal trajectories and the candidate trajectories according to the trained machine learning algorithm model, and determine whether the abnormal trajectories and the candidate trajectories in the same time window match according to the similarity of the characteristics of the abnormal trajectories and the candidate trajectories.

7. The method for vehicle trajectory tracking according to claim 5, characterized in that, The step of determining abnormal trajectories includes: Regard the trajectories that disappear midway and the trajectories that newly appear midway in the road segment as the abnormal trajectories in the road segment.

8. A system for vehicle trajectory tracking, characterized in that, It includes: A merging module configured to preliminarily merge the trajectories corresponding to multiple IDs assigned to the same vehicle to obtain the initial trajectory set of all vehicles; A grouping module configured to group the initial trajectory set by road segments and perform trajectory fitting on the target vehicles in each road segment; A matching module, configured to determine an abnormal trajectory based on a fitting result and match the abnormal trajectory with candidate trajectories other than the initial trajectory set in the same road segment; And An execution module, configured to, in response to a successful match, fuse the successfully matched candidate trajectory with the abnormal trajectory to form a trajectory of the road segment, and combine the trajectories of each road segment to form a final trajectory.

9. An electronic device, characterized in that, Comprising: At least one processor; And A memory storing computer instructions executable on the processor, and when the instructions are executed by the processor, the steps of the method according to any one of claims 1-7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1-7 are implemented.

Images