Vehicle Recognition Method, Device, Electronic Device and Storage Medium across Sensing Devices

Abstract

The present application provides a vehicle recognition method, device, electronic device and storage medium across sensing devices, which relates to the technical fields of cloud computing, artificial intelligence, and image recognition. The method includes: determining a target vehicle within the sensing range of the current sensing device, and determining a plurality of candidate vehicles that match the characteristics of the target vehicle among a plurality of vehicles within the sensing range of the previous sensing device adjacent to the current sensing device; based on the trajectory behavior information respectively corresponding to the target vehicle and the plurality of candidate vehicles, determining a matching vehicle that matches the target vehicle among the plurality of candidate vehicles; the trajectory behavior information includes attribute information of the position change behavior of the vehicle in the lane. In the technical solution of the present application, candidate vehicles are determined through feature matching, and a matching vehicle of the target vehicle is determined among the candidate vehicles based on the trajectory behavior information, which can improve the accuracy of vehicle recognition across sensing devices.

Description

Technical Field

[0001] The present application relates to the field of cloud computing technology, and in particular, to a vehicle recognition method, device, electronic device, and storage medium for cross-sensing devices. Background Art

[0002] Cross-sensing device target recognition refers to the ability to determine that the same target is still the same target after passing through different sensing devices, so as to complete the recognition task of the same target under different sensing devices. Cross-sensing device target recognition can be applied to application scenarios such as target trajectory restoration, target retrieval, and digital parallel worlds.

[0003] Currently, in the scenario of cross-sensing device vehicle recognition, it is easy to have recognition errors when the same vehicle model or similar vehicles cross-sensing devices, resulting in an increase in the difficulty of cross-sensing device recognition and a decrease in accuracy. Summary of the Invention

[0004] Embodiments of the present application provide a vehicle recognition method, device, electronic device, and storage medium for cross-sensing devices to improve the accuracy of cross-sensing device vehicle recognition.

[0005] In a first aspect, embodiments of the present application provide a vehicle recognition method for cross-sensing devices, including:

[0006] Determine a target vehicle within the sensing range of the current sensing device, and determine multiple candidate vehicles that match the characteristics of the target vehicle among multiple vehicles within the sensing range of the previous sensing device adjacent to the current sensing device;

[0007] Based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively, determine a matching vehicle that matches the target vehicle among the multiple candidate vehicles; the trajectory behavior information includes attribute information of the position change behavior of the vehicle in the lane.

[0008] In a second aspect, embodiments of the present application provide a vehicle recognition device for cross-sensing devices, including:

[0009] A candidate vehicle determination module, configured to determine a target vehicle within the sensing range of the current sensing device, and determine multiple candidate vehicles that match the characteristics of the target vehicle among multiple vehicles within the sensing range of the previous sensing device adjacent to the current sensing device;

[0010] A matching vehicle determination module, configured to determine a matching vehicle that matches the target vehicle among the multiple candidate vehicles based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively; the trajectory behavior information includes attribute information of the position change behavior of the vehicle in the lane.

[0011] In a third aspect, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory. When the processor executes the computer program, the method provided in any embodiment of the present application is implemented.

[0012] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the method provided in any embodiment of the present application is implemented.

[0013] Compared with the prior art, the present application has the following advantages:

[0014] The vehicle recognition method, device, electronic device, and storage medium across sensing devices provided in the embodiments of the present application first determine a plurality of candidate vehicles that match the target vehicle features, and then determine a matching vehicle that matches the target vehicle among the plurality of candidate vehicles. By feature matching to determine candidate vehicles and based on trajectory behavior information to determine the matching vehicle of the target vehicle among the candidate vehicles, the accuracy of vehicle recognition across sensing devices can be improved.

[0015] The above summary is only for the purpose of the specification and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the present application will be readily apparent by reference to the drawings and the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In the drawings, unless otherwise specified, the same reference numerals throughout the several views denote the same or similar components or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some embodiments disclosed in accordance with the present application and should not be regarded as limiting the scope of the present application.

[0017] Figure 1 It is a schematic diagram of an application scenario of the vehicle recognition method across sensing devices provided in an embodiment of the present application;

[0018] Figure 2 It is a flowchart of the vehicle recognition method across sensing devices provided in an embodiment of the present application;

[0019] Figure 3 It is a flowchart of the vehicle recognition method across sensing devices provided in an embodiment of the present application;

[0020] Figure 4 It is a schematic diagram of a scenario of different trajectory behaviors of the same vehicle model provided in an embodiment of the present application;

[0021] Figure 5 It is a flowchart of matching for the same vehicle model provided in an embodiment of the present application;

[0022] Figure 6 Schematic diagram of a vehicle recognition device across sensing devices provided by an embodiment of the present application;

[0023] Figure 7 Block diagram of an electronic device for implementing the embodiment of the present application. Specific embodiments

[0024] In the following, only some exemplary embodiments are briefly described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the present application. Therefore, the drawings and descriptions are considered to be exemplary in nature rather than restrictive.

[0025] To facilitate the understanding of the technical solutions of the embodiments of the present application, the related technologies of the embodiments of the present application are described below. The following related technologies can be arbitrarily combined with the technical solutions of the embodiments of the present application as optional solutions, and they all fall within the protection scope of the embodiments of the present application.

[0026] To more clearly show the vehicle recognition method across sensing devices provided in the embodiments of the present application, the application scenarios that can be used to implement this method are introduced first.

[0027] The technical solutions of the present application can be applied to application scenarios such as vehicle tracking, vehicle retrieval, vehicle trajectory restoration in the real world and the digital parallel world. In the technical solutions of the present application, the sensing device includes a device that can acquire an image of a target object. For example, a camera, a vision sensor, etc. In this embodiment, the camera is taken as an example for introduction. Figure 1 Schematic diagram of the application scenario of the vehicle recognition method across cameras provided by an embodiment of the present application. As Figure 1 shown, multiple cameras are set on the road to collect vehicle videos, including: Camera A and Camera B. The collection range of Camera A is the visible area A as shown in Figure 1 ; the collection range of Camera B is as shown in Figure 1Visible area B shown in [figure]; there is a blind area between visible area A and visible area B. After the vehicle enters the blind area, neither camera A nor camera B can collect the vehicle video. There are three lanes in the road: lane A, lane B, and lane C. Vehicles A, B, and C first enter the collection range of camera A, then enter the blind area, and after exiting the blind area, enter the collection range of camera B. Among them, vehicle A changed lanes during driving, changing from lane A to lane B; vehicle B changed from lane A to lane B and then back to lane A; vehicle C changed from lane B to lane C. Among them, the trigger line is a virtual detection line within the collection range of each camera. When the vehicle passes the trigger line corresponding to each camera, the feature information of the vehicle is extracted. The trigger line is set at a better visual position close to the camera to extract better vehicle features. In this embodiment, the current camera is camera B, and the previous camera adjacent to the current camera is camera A. For vehicle A within the collection range of camera B, it is currently impossible to determine which vehicle within the collection range of camera A is the same vehicle as vehicle A. When multiple vehicles pass the trigger line corresponding to camera A, the feature information of multiple vehicles is obtained, and multiple candidate vehicles that match the target vehicle features are determined among the multiple vehicles: vehicle A, vehicle B, and vehicle C. Based on the trajectory behavior information corresponding to the target vehicle and vehicles A, B, and C respectively, the matching vehicle that matches the target vehicle is determined among vehicles A, B, and C, thereby realizing vehicle recognition across cameras.

[0028] An embodiment of the present application provides a vehicle recognition method across sensing devices. Figure 2 It is a flowchart of the vehicle recognition method across sensing devices according to an embodiment of the present application. This method can be applied to a vehicle recognition device across sensing devices, and this device can be deployed in a user terminal, a server, or other processing devices. In some possible implementation manners, this method can also be implemented by a processor calling computer-readable instructions stored in a memory. As Figure 2 shown, this method includes:

[0029] Step S201, determine the target vehicle within the sensing range of the current sensing device, and determine multiple candidate vehicles that match the target vehicle features among multiple vehicles within the sensing range of the previous sensing device adjacent to the current sensing device;

[0030] In this embodiment, the execution subject may be a server. The sensing device includes a device that can obtain an image of a target object. For example, a camera, a vision sensor, etc. In this embodiment, the camera is taken as an example for introduction. The server obtains the video of the vehicle driving on the road through the camera. Multiple cameras can be set on the road. The matching vehicle of the target vehicle can be obtained through the videos of the vehicle collected by multiple cameras, and then the driving trajectory of the target vehicle can be obtained.

[0031] Obtain the feature information of the target vehicle within the acquisition range of the current camera, as well as the feature information of multiple vehicles within the acquisition range of the previous camera. Multiple candidate vehicles can be obtained through feature matching. Feature matching can be calculating the similarity between the feature information of the target vehicle and the feature information of multiple vehicles, and obtaining multiple vehicles with higher similarity to the target vehicle as candidate vehicles according to the order of similarity from large to small.

[0032] Step S202: Based on the trajectory behavior information corresponding to the target vehicle and multiple candidate vehicles respectively, determine the matching vehicle that matches the target vehicle among the multiple candidate vehicles.

[0033] Among them, the trajectory behavior information includes the attribute information of the position change behavior of the vehicle in the lane. The position change behavior of the vehicle in the lane may include, but is not limited to, lane keeping, lane changing, overtaking by changing lanes, returning to the original lane after overtaking by changing lanes, and overtaking in different lanes. The attribute information of the position change behavior of the vehicle in the lane may include, but is not limited to, time, lane identification, and vehicle speed. Based on the trajectory behavior information corresponding to the target vehicle and multiple candidate vehicles respectively, obtain the matching degrees of the multiple candidate vehicles corresponding to the target vehicle respectively, and determine the matching vehicle of the target vehicle according to the multiple matching degrees.

[0034] The vehicle recognition method across sensing devices provided by the embodiments of the present application first determines multiple candidate vehicles that match the features of the target vehicle, and then determines the matching vehicle that matches the target vehicle among the multiple candidate vehicles. By determining candidate vehicles through feature matching and determining the matching vehicle of the target vehicle among the candidate vehicles based on the trajectory behavior information, the accuracy of vehicle recognition across sensing devices can be improved.

[0035] For ease of understanding, the sensing devices in the following embodiments are described by taking cameras as examples. Among them, the sensing range of the sensing device is the acquisition range of the camera.

[0036] Among them, the specific implementation manner of determining the vehicle that matches the target vehicle is as follows in the embodiments:

[0037] In a possible implementation manner, in step S202, based on the trajectory behavior information corresponding to the target vehicle and multiple candidate vehicles respectively, determining the matching vehicle that matches the target vehicle among the multiple candidate vehicles includes:

[0038] Step S2021: Based on the trajectory behavior information corresponding to the target vehicle and multiple candidate vehicles respectively, determine the time weight and lane weight of the multiple candidate vehicles;

[0039] Step S2022: Based on the time weight and lane weight of the multiple candidate vehicles, determine the matching vehicle that matches the target vehicle among the multiple candidate vehicles.

[0040] Among them, the trajectory behavior information includes the attribute information of the position change behavior of the vehicle in the lane. According to the attribute information, the time weight and lane weight of each candidate vehicle among multiple candidate vehicles can be determined. For any candidate vehicle, the time weight is used to characterize the probability that the candidate vehicle appears earlier than other candidate vehicles in the current camera. The higher the probability that the candidate vehicle appears earlier in the current camera, the higher the matching degree with the target vehicle in the time dimension. The lane weight is used to characterize the probability that the lanes where multiple candidate vehicles are located are the same as the lane where the target vehicle is located. The higher the probability that the lane where the candidate vehicle is located is the same as the lane where the target vehicle is located, the higher the matching degree with the target vehicle in the lane dimension. Therefore, among multiple candidate vehicles, the candidate vehicle with a larger time weight and lane weight has a higher matching degree with the target vehicle.

[0041] Optionally, the time weight and lane weight of the candidate vehicle can be multiplied, and the candidate vehicle with the largest product value in the calculation result is used as the matching vehicle that matches the target vehicle.

[0042] In this embodiment, on the basis of feature matching, based on the trajectory behavior information, the matching vehicle of the target vehicle is further determined from the candidate vehicles in the time dimension and the lane dimension, which can further improve the accuracy of vehicle matching.

[0043] In a possible implementation manner, step S2021, based on the trajectory behavior information corresponding to the target vehicle and multiple candidate vehicles respectively, determining the time weight and lane weight of the multiple candidate vehicles includes:

[0044] Performing a first difference calculation based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when multiple candidate vehicles disappear from the sensing range of the previous sensing device to obtain the time weights of the multiple candidate vehicles;

[0045] Performing a second difference calculation based on the lane identifier of the target vehicle passing through the current sensing device and the lane identifier when multiple candidate vehicles disappear from the sensing range of the previous sensing device to obtain the lane weights of the multiple candidate vehicles.

[0046] Among them, the trajectory behavior information corresponding to the target vehicle includes the time and lane identifier when the target vehicle passes through the acquisition range of the current camera. The trajectory behavior information corresponding to multiple candidate vehicles includes the time and lane identifier when multiple candidate vehicles disappear from the acquisition range of the previous camera.

[0047] The time when the target vehicle passes through the acquisition range of the current camera can be the time when the vehicle arrives at the virtual detection line corresponding to the current camera. The virtual detection line can be used as a unified reference position for the vehicle to enter the camera acquisition range. The lane identifier of the target vehicle passing through the acquisition range of the current camera can be the lane number where the target vehicle is located when passing through the acquisition range of the current camera. It should be noted that the time when the target vehicle passes through the acquisition range of the current camera is not limited to the time of arriving at the position of the virtual detection line, and can also be the time at other positions within the camera acquisition range. This application does not make any restrictions on this. Based on the time when the target vehicle passes through the acquisition range of the current camera and the time when multiple candidate vehicles disappear from the acquisition range of the previous camera, by performing the first difference calculation, the time weights of multiple candidate vehicles can be obtained. The time weights can measure the matching degree between the candidate vehicles and the target vehicle in the time dimension. The earlier the time when a candidate vehicle disappears from the acquisition range of the previous camera, the higher the matching degree between the candidate vehicle and the target vehicle in the time dimension.

[0048] Based on the lane identifier of the target vehicle passing through the current camera and the lane identifiers of multiple candidate vehicles when they disappear from the acquisition range of the previous camera, by performing the second difference calculation, the lane weights of multiple candidate vehicles can be obtained. The lane weights can measure the lane matching degree between each candidate vehicle and the target vehicle. The higher the matching degree between the lane identifier of the candidate vehicle when it disappears from the acquisition range of the previous camera and the lane identifier of the lane where the target vehicle is currently located, the higher the probability that the candidate vehicle and the target vehicle are in the same lane.

[0049] There may be a blind area between the acquisition range of the previous camera and the acquisition range of the current camera. Within the blind area, neither the previous camera nor the current camera can collect the video of the vehicle. After a candidate vehicle disappears from the acquisition range of the previous camera and enters the blind area, since the blind area is relatively small, the possibility of the candidate vehicle changing lanes in the blind area is relatively small. Therefore, the higher the matching degree between the lane identifier of the candidate vehicle when it enters the blind area and the lane identifier of the lane where the target vehicle is currently located, the higher the matching degree between the candidate vehicle and the target vehicle in the lane dimension.

[0050] Among them, the specific implementation method for determining the time weight of the candidate vehicle is shown in the following embodiments:

[0051] In a possible implementation manner, based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when multiple candidate vehicles disappear from the sensing range of the previous sensing device, perform the first difference calculation to obtain the time weights of multiple candidate vehicles, including:

[0052] Based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when multiple candidate vehicles disappear from the sensing range of the previous sensing device, determine the cross-sensing device times of multiple candidate vehicles;

[0053] Determine the time weights of multiple candidate vehicles according to the cross-sensing device times of the multiple candidate vehicles.

[0054] Among them, calculate the time difference between the time when the target vehicle passes through the acquisition range of the current camera and the time when the multiple candidate vehicles disappear from the acquisition range of the previous camera respectively, as the cross-camera time corresponding to each candidate vehicle. Determine the time weights of each candidate vehicle according to the cross-camera times of the multiple candidate vehicles.

[0055] Optionally, for each candidate vehicle, use the ratio of the cross-camera time of the candidate vehicle to the maximum value among the cross-camera times of all candidate vehicles as the time weight of the candidate vehicle.

[0056] In a possible implementation manner, determining a matching vehicle that matches the target vehicle among multiple candidate vehicles based on the time weights and lane weights of the multiple candidate vehicles includes:

[0057] Determine the feature weights of multiple candidate vehicles;

[0058] Determine the matching vehicle of the target vehicle among multiple candidate vehicles according to the time weights, lane weights, and feature weights of the multiple candidate vehicles.

[0059] In practical applications, the feature weights of multiple candidate vehicles can be determined based on the feature information of the target vehicle and the feature information of the multiple candidate vehicles. According to the three dimensions of the time weights, lane weights, and feature weights of the multiple candidate vehicles, determine the matching vehicle of the target vehicle among the multiple candidate vehicles. Optionally, for any candidate vehicle, the time weight, lane weight, and feature weight of the candidate vehicle can be multiplied, and the obtained calculation result can be used as the comprehensive weight of the candidate vehicle. The candidate vehicle with the largest comprehensive weight is determined as the matching vehicle of the target vehicle, and the global identifier of the candidate vehicle is used as the identifier of the target vehicle, so as to achieve vehicle identification.

[0060] In a possible implementation manner, determining the feature weights of multiple candidate vehicles includes:

[0061] Based on the feature information of the target vehicle and the feature information of the multiple candidate vehicles, determine the similarity between the target vehicle and the multiple candidate vehicles, and determine the similarities corresponding to the multiple candidate vehicles as the feature weights; the feature information is extracted when the vehicle passes through the virtual detection line corresponding to the first sensing device.

[0062] In practical applications, the similarity between the target vehicle and multiple candidate vehicles can be calculated based on the feature information of the target vehicle and the feature information of the multiple candidate vehicles. For any candidate vehicle, the similarity between the candidate vehicle and the target vehicle is used as the feature weight of the candidate vehicle. Among them, the feature information includes, but is not limited to, vehicle features extracted from video frames collected by a convolutional neural network model from a camera, vehicle model features, color features, etc. The timing of extracting the feature information of the vehicle can be when the vehicle passes through the trip wire corresponding to each of the multiple cameras set on the road. The vehicle features can be extracted only once at each trip wire, because the trip wire is a position with better camera vision. Extracting the vehicle features once at each trip wire does not require multiple extractions, which can improve the calculation efficiency while ensuring accuracy.

[0063] In one possible implementation, before determining the matching vehicle that matches the target vehicle among the multiple candidate vehicles based on the trajectory behavior information respectively corresponding to the target vehicle and the multiple candidate vehicles, it further includes:

[0064] Determine that the difference between the two largest similarities in the similarities between the target vehicle and the multiple candidate vehicles is less than a preset threshold.

[0065] In practical applications, after determining the multiple candidate vehicles, calculate the difference between the two largest similarities in the similarities corresponding to the candidate vehicles. If the difference is less than the preset threshold, it indicates that the models of the multiple candidate vehicles and the target vehicle may be the same or similar. Then, based on the trajectory behavior information respectively corresponding to the target vehicle and the multiple candidate vehicles, determine the matching vehicle that matches the target vehicle among the multiple candidate vehicles.

[0066] In one possible implementation, the method further includes:

[0067] In the case where the difference between the two largest similarities in the similarities between the target vehicle and the multiple candidate vehicles is greater than the preset threshold, determine the candidate vehicle with the highest similarity as the matching vehicle of the target vehicle.

[0068] In practical applications, after determining the multiple candidate vehicles, calculate the difference between the two largest similarities in the similarities corresponding to the candidate vehicles. If the difference between the two largest similarities in the similarities between the target vehicle and the multiple candidate vehicles is greater than the preset threshold, it indicates that, except for the candidate vehicle with the highest similarity, the models of the other candidate vehicles and the target vehicle may not be the same or similar. Then, use the candidate vehicle with the highest similarity as the matching vehicle of the target vehicle. Optionally, it can also be determined whether the appearance features of the matching vehicle and the target vehicle match to further determine whether the matching vehicle and the target vehicle are the same vehicle. Among them, the appearance features include vehicle model, color, etc.

[0069] In a possible implementation, among multiple vehicles within the sensing range of the previous sensing device adjacent to the current sensing device, determining multiple candidate vehicles that match the target vehicle features includes:

[0070] Determining vehicles that meet the time constraint condition and the space constraint condition among the multiple vehicles, and among the vehicles that meet the time constraint condition and the space constraint condition, determining multiple candidate vehicles that match the target vehicle;

[0071] Among them, the space constraint condition is determined based on the positions of the multiple vehicles, the distances between the target vehicle and the multiple vehicles, and the distance between the current sensing device and the previous sensing device; the time constraint condition is determined based on the times when the target vehicle and the multiple vehicles respectively pass through the sensing ranges of the current sensing device and the previous sensing device.

[0072] Optionally, determining whether the space constraint condition is met includes at least one of the following: determining whether the multiple vehicles are within the lane, determining whether the difference between the vehicle distance and the camera distance is less than a preset distance threshold, where the vehicle distance is the distance between the target vehicle and any one of the multiple vehicles, and the camera distance is the distance between the current camera and the previous camera.

[0073] Determining whether the time constraint condition is met includes at least one of the following: determining whether the time difference between the time when any one of the multiple vehicles passes through the trip wire corresponding to the previous camera and the time when the target vehicle passes through the trip wire corresponding to the current camera is less than a preset time threshold; the time difference calculated from the time when any one of the vehicles passes through the trip wire corresponding to the previous camera and the time when the target vehicle passes through the trip wire corresponding to the current camera, the first estimated time when any one of the vehicles passes through the two trip wires, the second estimated time when the target vehicle passes through the two trip wires, and performing a difference calculation on the time difference with the first estimated time and the second estimated time respectively, and determining whether the smaller calculated time difference is less than a preset time threshold. Among them, for any one vehicle, the first estimated time can be calculated by the distance that the vehicle passes through the two trip wires and the current speed of the vehicle; the second estimated time can be calculated by the distance that the target vehicle passes through the two trip wires and the current speed of the target vehicle.

[0074] Among them, if the judgment results of the above time constraint condition and space constraint condition are yes, it indicates that the time constraint condition and the space constraint condition are met.

[0075] In a possible implementation, before determining the matching vehicle that matches the target vehicle among the multiple candidate vehicles based on the trajectory behavior information respectively corresponding to the target vehicle and the multiple candidate vehicles, the method further includes:

[0076] Filtering out candidate vehicles that appear within the sensing range of the previous sensing device and have not disappeared within the acquisition duration range.

[0077] In practical applications, before determining a matching vehicle that matches the target vehicle from multiple candidate vehicles based on the trajectory behavior information corresponding to the target vehicle and each of the multiple candidate vehicles, real-time vehicles in the previous camera are filtered out. If a vehicle appears within the acquisition range of the previous camera and does not disappear within the acquisition duration range, that is, it does not enter the blind area between the two cameras, then it will not appear in the current camera and thus will not be a vehicle that matches the target vehicle. Therefore, this type of candidate vehicle is filtered out and not matched, which also has the effect of saving computing resources.

[0078] In a possible implementation manner, the method further includes:

[0079] In a case where there is a blind area between the sensing range of the previous sensing device and the sensing range of the current sensing device, based on the trajectory behavior information corresponding to the target vehicle and the matching vehicle respectively, a virtual driving trajectory of a virtual target vehicle corresponding to the target vehicle is determined in a virtual environment, and the virtual driving trajectory includes the driving trajectory of the virtual target vehicle in the virtual blind area corresponding to the blind area.

[0080] In practical applications, in a case where there is a blind area between the acquisition range of the previous camera and the acquisition range of the current camera, the virtual driving trajectory of the virtual target vehicle corresponding to the target vehicle can be simulated in a virtual environment according to the time, lane, speed, etc. of the matching vehicle within the acquisition range of the previous camera, and the time, lane, speed, etc. of the target vehicle within the acquisition range of the current camera. The starting point of the virtual driving trajectory can be determined according to the position where the target vehicle appears collected by the first camera, and the end point of the virtual driving trajectory point can be determined according to the position where the target vehicle appears collected by the current camera. The virtual driving trajectory can be updated in real time as the target vehicle continues to drive; it is also possible to generate a complete virtual driving trajectory of the target vehicle during the entire journey after collecting the trajectory behavior information of the target vehicle and the matching vehicle in the acquisition range of each camera. Among them, the virtual driving trajectory includes the driving trajectory of the simulated virtual target vehicle in the virtual blind area. In this embodiment, the restoration of the vehicle driving trajectory in the virtual environment can be realized.

[0081] Figure 3 This is a flowchart of a cross-sensing device vehicle recognition method provided by an embodiment of the present application. The scenario of this embodiment is a highway, and multiple cameras are set on the highway. The cross-sensing device vehicle recognition method in this embodiment includes the following steps:

[0082] (1) Different cameras simultaneously perform vehicle detection and recognition.

[0083] Specifically, multiple cameras collect videos of vehicles driving on the highway, and vehicle detection and recognition are performed based on the collected videos.

[0084] (2) Each off-road camera extracts and caches vehicle features separately.

[0085] Specifically, when the vehicle reaches the tripwire corresponding to the first camera, the features of the vehicle are extracted and cached through the convolutional neural network model. At the same time, the time when the vehicle passes the tripwire, the lane it belongs to, the vehicle speed and other attribute information are recorded.

[0086] (3) The next vehicle calculates the feature similarity with all the vehicles that meet the spatio-temporal constraint conditions of the previous vehicle, and selects the top K vehicles (K is a positive integer and K≥2) with the highest similarity to obtain the matching candidate set G K .

[0087] Among them, the next vehicle is the vehicle within the acquisition range of the next camera, and this vehicle is used as the target vehicle. All the vehicles that meet the spatio-temporal constraint conditions of the previous vehicle are multiple vehicles within the acquisition range of the previous camera. The spatio-temporal constraint conditions include time constraint conditions and space constraint conditions. Calculate the similarity between the target vehicle and multiple vehicles within the acquisition range of the previous camera, perform feature matching based on multiple similarities, obtain multiple vehicles, and select the top K vehicles with the highest similarity as candidate vehicles from them to form the matching candidate set G K .

[0088] (4) Determine whether the difference between the first highest similarity and the second highest similarity is greater than the threshold T.

[0089] Select the two highest similarities among the similarities corresponding to each candidate vehicle in the candidate set G K , the first highest similarity and the second highest similarity, calculate their difference. If the difference is less than the preset threshold T, it indicates that the models of multiple candidate vehicles may be the same or similar to the target vehicle, then enter the same model matching module, perform matching through the same model matching module, and based on the trajectory behavior information corresponding to the target vehicle and multiple candidate vehicles respectively, determine the matching vehicle that matches the target vehicle among multiple candidate vehicles.

[0090] If the difference is greater than the preset threshold T, it indicates that except for the candidate vehicle with the highest similarity, the models of other candidate vehicles may be different or not similar to the target vehicle. Therefore, the candidate vehicle with the highest similarity is determined as the matching vehicle of the target vehicle.

[0091] In summary, through the processing process of the above steps, the matching vehicle of the target vehicle can be obtained, and the global identifier of this matching vehicle is used as the identifier of the target vehicle to achieve cross-camera recognition of the target vehicle.

[0092] For vehicles of the same model, their attribute features such as vehicle appearance and color are the same, and the convolutional neural network feature similarity is close. It is difficult to distinguish vehicles based on these features. However, their trajectory behavior information is different, and vehicles of the same model can be distinguished based on their different trajectory behavior information. The different trajectory behavior scenario graphs of vehicles of the same model are as shown in Figure 4 as follows.

[0093] Multiple cameras are set up on the road to collect vehicle videos, including: Camera A and Camera B. The collection range of Camera A is the visible area A as shown in Figure 4 ; the collection range of Camera B is the visible area B as shown in Figure 1 ; there is a blind area between the visible area A and the visible area B. After the vehicle drives into the blind area, neither Camera A nor Camera B can collect the vehicle video. There are three lanes on the road: Lane A, Lane B, and Lane C. Vehicles A, B, and C of the same model (the same model A, the same model B, and the same model C shown in the figure) first drive into the collection range of Camera A, then drive into the blind area, and after driving out of the blind area, drive into the collection range of Camera B.

[0094] Vehicles A, B, and C may all change lanes during driving. Take any vehicle within the collection range of Camera B as the target vehicle, determine the candidate vehicles with feature matching among the vehicles within the collection range of Camera A, calculate the time weight and lane weight corresponding to each candidate vehicle according to the trajectory behavior information corresponding to the target vehicle and the candidate vehicles, take the similarity between the candidate vehicle and the target vehicle as the feature weight. For each candidate vehicle, multiply the time weight, lane weight, and feature weight, and take the candidate vehicle with the largest product as the matching vehicle of the target vehicle. For example, if the vehicle in Lane A within the current camera collection range is the target vehicle and the final matching vehicle is Vehicle C, then assign the global identifier of Vehicle C to the target vehicle.

[0095] Among them, the trajectory behavior information includes: and represents the time when the k-th vehicle within the collection range of the camera numbered n (Camera A in this embodiment) passes the detection line, represents the lane number where the k-th vehicle within the collection range of the camera numbered n passes the detection line, represents the time when the k-th vehicle within the collection range of the camera numbered n disappears, represents the lane number where the k-th vehicle within the collection range of the camera numbered n disappears, represents the time when the k-th vehicle within the collection range of the camera numbered n + 1 (Camera B in this embodiment) passes the detection line, It represents the lane number where the k-th vehicle within the acquisition range of the camera numbered n + 1 passes through the trigger line. Here, i represents the input, and o represents the output.

[0096] Figure 5 It is a flowchart for matching the same vehicle models provided by an embodiment of this application. The cross-camera vehicle recognition method in this embodiment includes the following steps:

[0097] (1) Select the vehicle with the maximum similarity and the vehicles with similarity less than the threshold T to the vehicle with the maximum similarity from the set G of K candidate vehicles K to form a candidate set G containing K' vehicles. K′ .

[0098] Among them, K takes the value of 3, and 2 ≤ K' ≤ K. The vehicles in the set G K′ are a set of multiple vehicles of the same model or similar models.

[0099] (2) Filter out the real-time vehicles still existing in their respective cameras from G K′ to obtain a final candidate set G containing K'' vehicles. K″ .

[0100] Among them, the real-time vehicles still existing in their respective cameras are those that appeared in the previous camera and did not disappear within the acquisition duration range, so they will not appear in the current camera and do not need to be matched, and such vehicles are filtered out. Among them, 0 ≤ K'' ≤ K'. When K'' = 0, all the vehicles in the previous camera have not disappeared yet, so a global identifier is configured for the vehicles within the acquisition range of the current camera, and the matching process ends. The set G K″ is a set of multiple finally determined candidate vehicles.

[0101] (3) Calculate the time weight of each vehicle in G K″ with the vehicle to be matched on the next road.

[0102] The vehicle to be matched on the next road is the target vehicle within the acquisition range of the current camera. Based on the time when the target vehicle passes through the acquisition range of the current camera and the time when multiple candidate vehicles disappear from the acquisition range of the previous camera, the cross-camera time of multiple candidate vehicles is determined; according to the cross-camera time of multiple candidate vehicles, the time weight of multiple candidate vehicles is determined.

[0103] The cross-camera time is calculated using the following formula:

[0104]

[0105] Among them, Indicates the time when the k-th vehicle disappears from the n-th camera and passes the trip wire corresponding to the (n + 1)-th camera, i.e., the cross-camera time; t (n+1)i Indicates the time when the target vehicle passes the trip wire corresponding to the (n + 1)-th camera (the current camera); Indicates the time when the k-th vehicle disappears from the n-th camera (the previous camera); n o Indicates the position where the vehicle disappears from the camera, n i Indicates the position where the vehicle passes the trip wire corresponding to the camera. k represents the k-th vehicle in the set G K″ where 0 ≤ k ≤ K″.

[0106] The time weight is calculated using the following formula:

[0107]

[0108] where, represents the time weight of the k-th candidate vehicle. The earlier the vehicle disappears from the n-th camera, i.e., the smaller it is, the greater the probability of its appearance from the (n + 1)-th camera, i.e., the greater it is.

[0109] The candidate vehicle with time weight is represented using the following formula:

[0110]

[0111] where, represents the k-th vehicle with time weight; represents the k-th vehicle; t represents calculating the matching vehicle in the time dimension.

[0112] (4) Calculate the lane weight of each vehicle in G K″ with the vehicle to be matched on the next road.

[0113] Based on the lane identification of the target vehicle passing the current camera and the lane identifications of multiple candidate vehicles when they disappear from the acquisition range of the previous camera, the lane weights of multiple candidate vehicles are obtained.

[0114] The lane weight is calculated using the following formula:

[0115]

[0116] where, represents the lane weight of the k-th candidate vehicle, represents the lane number of the k-th vehicle when it disappears from the n-th camera; l (n+1)iIt represents the lane number when the target vehicle within the acquisition range of the (n + 1)-th camera passes through the trigger line; L represents the maximum lane number. The greater the probability that the lane where the k-th candidate vehicle disappears from the n-th camera is close to the lane where the target vehicle passes through the trigger line at the (n + 1)-th camera, that is the greater.

[0117] The candidate vehicle with lane weight is represented by the following formula:

[0118]

[0119] where represents the k-th vehicle with lane weight; represents the k-th vehicle; l represents calculating the matching vehicle in the lane dimension.

[0120] (5) Multiply the time weight and lane weight of the same lane, and then multiply by the vehicle feature similarity to obtain the final weight of the candidate vehicle.

[0121] where the feature similarity is the feature weight, that is, the similarity between the candidate vehicle and the target vehicle.

[0122] The candidate vehicle with feature weight is represented by the following formula:

[0123]

[0124] where represents the k-th candidate vehicle with feature weight; represents the feature weight of the k-th candidate vehicle; represents the k-th vehicle; s represents calculating the matching vehicle in the feature dimension.

[0125] The final weight of the candidate vehicle can be calculated by the following formula:

[0126]

[0127] where represents the final weight of the k-th candidate vehicle; represents the time weight of the k-th candidate vehicle; represents the lane weight of the k-th candidate vehicle; represents the feature weight of the k-th candidate vehicle.

[0128] The candidate vehicle with the final weight is represented by the following formula:

[0129]

[0130] where represents the k-th candidate vehicle with the final weight; Denote the k-th candidate vehicle.

[0131] (6) Select the vehicle with the largest weight as the final matching vehicle.

[0132] Select the maximum weight according to formula (8) The corresponding vehicle As the final matching vehicle, complete the cross-camera tracking of the same vehicle model.

[0133] Corresponding to the application scenario and method of the method provided in the embodiments of the present application, the embodiments of the present application also provide a vehicle recognition device across sensing devices. As Figure 6 shown, the vehicle recognition device across sensing devices may include:

[0134] A candidate vehicle determination module 601, configured to determine a target vehicle within the sensing range of the current sensing device, and determine multiple candidate vehicles that match the characteristics of the target vehicle among multiple vehicles within the sensing range of the previous sensing device adjacent to the current sensing device;

[0135] A matching vehicle determination module 602, configured to determine a matching vehicle that matches the target vehicle among multiple candidate vehicles based on the trajectory behavior information respectively corresponding to the target vehicle and the multiple candidate vehicles; the trajectory behavior information includes attribute information on the position change behavior of the vehicle in the lane.

[0136] The vehicle recognition device across sensing devices provided by the embodiments of the present application first determines multiple candidate vehicles that match the characteristics of the target vehicle, and then determines a matching vehicle that matches the target vehicle among the multiple candidate vehicles. By determining candidate vehicles through feature matching and determining the matching vehicle of the target vehicle among the candidate vehicles based on trajectory behavior information, the accuracy of vehicle recognition across sensing devices can be improved.

[0137] In a possible implementation manner, the matching vehicle determination module 602 includes a weight determination unit and a vehicle determination unit, and specifically is used for:

[0138] The weight determination unit is configured to determine the time weight and lane weight of multiple candidate vehicles based on the trajectory behavior information respectively corresponding to the target vehicle and the multiple candidate vehicles;

[0139] The vehicle determination unit is configured to determine a matching vehicle that matches the target vehicle among multiple candidate vehicles based on the time weight and lane weight of the multiple candidate vehicles;

[0140] Among them, for any candidate vehicle, the time weight is used to characterize the probability that the candidate vehicle appears earlier in the current sensing device than other candidate vehicles;

[0141] The lane weight is used to characterize the probability that the lanes where multiple candidate vehicles are located are the same lane as the lane where the target vehicle is located.

[0142] In a possible implementation, a weight determination unit is configured to:

[0143] Perform a first difference calculation based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when multiple candidate vehicles disappear from the sensing range of the previous sensing device, to obtain the time weights of the multiple candidate vehicles;

[0144] Perform a second difference calculation based on the lane identifier of the target vehicle passing through the sensing range of the current sensing device and the lane identifier of the multiple candidate vehicles when they disappear from the sensing range of the previous sensing device, to obtain the lane weights of the multiple candidate vehicles;

[0145] Wherein, the trajectory behavior information corresponding to the target vehicle includes the time when the target vehicle passes through the sensing range of the current sensing device and the lane identifier; the trajectory behavior information corresponding to the multiple candidate vehicles includes the time when the multiple candidate vehicles disappear from the sensing range of the previous sensing device and the lane identifier.

[0146] In a possible implementation, when the weight determination unit performs a first difference calculation based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when multiple candidate vehicles disappear from the sensing range of the previous sensing device to obtain the time weights of the multiple candidate vehicles, it is configured to:

[0147] Determine the cross-sensing device times of the multiple candidate vehicles based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when the multiple candidate vehicles disappear from the sensing range of the previous sensing device;

[0148] Determine the time weights of the multiple candidate vehicles according to the cross-sensing device times of the multiple candidate vehicles.

[0149] In a possible implementation, a vehicle determination unit is configured to:

[0150] Determine the feature weights of the multiple candidate vehicles;

[0151] Determine the matching vehicle of the target vehicle among the multiple candidate vehicles according to the time weights, lane weights and feature weights of the multiple candidate vehicles.

[0152] In a possible implementation, when the vehicle determination unit determines the feature weights of the multiple candidate vehicles, it is configured to:

[0153] Determine the similarity between the target vehicle and the multiple candidate vehicles based on the feature information of the target vehicle and the feature information of the multiple candidate vehicles, and determine the similarities corresponding to the multiple candidate vehicles as the feature weights; the feature information is extracted when the vehicle passes through the virtual detection line corresponding to the first sensing device.

[0154] In a possible implementation, the device further includes a similarity comparison module, configured to:

[0155] Determine that the difference between the two largest similarities among the similarities between the target vehicle and multiple candidate vehicles is less than a preset threshold.

[0156] In a possible implementation, the device further includes a feature comparison module, configured to:

[0157] In the case where the difference between the two largest similarities among the similarities between the target vehicle and multiple candidate vehicles is greater than a preset threshold, determine the candidate vehicle with the highest similarity as the matching vehicle of the target vehicle.

[0158] In a possible implementation, the candidate vehicle determination module 601 is specifically configured to:

[0159] Determine vehicles that meet the time constraint condition and the space constraint condition among multiple vehicles, and determine multiple candidate vehicles that match the target vehicle among the vehicles that meet the time constraint condition and the space constraint condition;

[0160] Wherein, the space constraint condition is determined based on the positions of multiple vehicles, the distances between the target vehicle and multiple vehicles, and the distance between the current sensing device and the previous sensing device; the time constraint condition is determined based on the times when the target vehicle and multiple vehicles respectively pass through the sensing ranges of the current sensing device and the previous sensing device.

[0161] In a possible implementation, the candidate vehicle determination module 601 is further configured to:

[0162] Filter out candidate vehicles that appear within the sensing range of the previous sensing device and have not disappeared within the acquisition duration range.

[0163] In a possible implementation, the device further includes a trajectory generation module, configured to:

[0164] In the case where there is a blind area between the sensing range of the previous sensing device and the sensing range of the current sensing device, based on the respective trajectory behavior information of the target vehicle and the matching vehicle, determine the virtual driving trajectory of the virtual target vehicle corresponding to the target vehicle in the virtual environment, and the virtual driving trajectory includes the driving trajectory of the virtual target vehicle in the virtual blind area corresponding to the blind area.

[0165] For the functions of the modules in each device of the embodiments of the present application, reference may be made to the corresponding descriptions in the above methods, and they have corresponding beneficial effects, which will not be elaborated here.

[0166] Figure 7 It is a block diagram of an electronic device for implementing the embodiments of the present application. As Figure 7As shown in the figure, the electronic device includes: a memory 710 and a processor 720. The memory 710 stores a computer program that can run on the processor 720. When the processor 720 executes the computer program, the methods in the above embodiments are implemented. The number of the memory 710 and the processor 720 can be one or more.

[0167] The electronic device further includes:

[0168] a communication interface 730, configured to communicate with external devices and perform data interaction and transmission.

[0169] If the memory 710, the processor 720, and the communication interface 730 are implemented independently, the memory 710, the processor 720, and the communication interface 730 can be interconnected through a bus and complete communication with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience of representation, Figure 7 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus.

[0170] Optionally, in specific implementation, if the memory 710, the processor 720, and the communication interface 730 are integrated on a chip, the memory 710, the processor 720, and the communication interface 730 can complete communication with each other through an internal interface.

[0171] An embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, the methods provided in the embodiments of the present application are implemented.

[0172] An embodiment of the present application further provides a chip, which includes a processor, configured to call and run instructions stored in a memory, so that a communication device installed with the chip executes the methods provided in the embodiments of the present application.

[0173] An embodiment of the present application further provides a chip, including: an input interface, an output interface, a processor, and a memory. The input interface, the output interface, the processor, and the memory are connected through an internal connection path. The processor is configured to execute code in the memory. When the code is executed, the processor is configured to execute the methods provided in the embodiments of the application.

[0174] It should be understood that the above-mentioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. It is worth noting that the processor can be a processor that supports the Advanced RISC Machines (ARM) architecture.

[0175] Furthermore, optionally, the above-mentioned memory can include a read-only memory and a random access memory, and can also include a non-volatile random access memory. The memory can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can include a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically EPROM (EEPROM), or a flash memory. The volatile memory can include a Random Access Memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available. For example, Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DR RAM).

[0176] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions according to the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium.

[0177] In the description of this specification, the descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0178] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" can explicitly or implicitly include at least one of these features. In the description of the present application, "a plurality of" means two or more unless otherwise specifically defined.

[0179] Any process or method description represented in the flowchart or described in other ways herein can be understood as representing a module, segment, or part of the code of executable instructions including one or more steps for implementing a specific logical function or process. And the scope of the preferred embodiments of the present application includes additional implementations, where the functions can be executed in a substantially simultaneous manner or in the reverse order according to the involved functions, rather than in the order shown or discussed.

[0180] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a sequenced list of executable instructions for implementing a logical function, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatus, or devices.

[0181] It should be understood that each part of the present application can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. All or part of the steps of the method in the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0182] In addition, each functional unit in various embodiments of the present application can be integrated into a processing module, or each unit can exist physically alone, or two or more units can be integrated into one module. The above integrated module can be implemented in the form of hardware or in the form of a software functional module. When the above integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The storage medium can be a read-only memory, a magnetic disk, an optical disk, etc.

[0183] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of various changes or substitutions, and these should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A vehicle recognition method across sensing devices, characterized in that The method includes: Determine a target vehicle within the sensing range of the current sensing device, and among multiple vehicles within the sensing range of the previous sensing device adjacent to the current sensing device, determine multiple candidate vehicles that match the characteristics of the target vehicle; Based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively, determine a matching vehicle that matches the target vehicle among the multiple candidate vehicles; the trajectory behavior information includes attribute information on the position change behavior of the vehicle in the lane; wherein, based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively, determining a matching vehicle that matches the target vehicle among the multiple candidate vehicles includes: based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively, determine the time weight and lane weight of the multiple candidate vehicles; based on the time weight and lane weight of the multiple candidate vehicles, determine a matching vehicle that matches the target vehicle among the multiple candidate vehicles; based on the time weight and lane weight of the multiple candidate vehicles, determining a matching vehicle that matches the target vehicle among the multiple candidate vehicles includes: determine the characteristic weight of the multiple candidate vehicles; according to the time weight, lane weight, and characteristic weight of the multiple candidate vehicles, determine the matching vehicle of the target vehicle among the multiple candidate vehicles; Based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively, determining the time weight and lane weight of the multiple candidate vehicles includes: Perform a first difference calculation based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when the multiple candidate vehicles disappear from the sensing range of the previous sensing device, to obtain the time weights of the multiple candidate vehicles; Perform a second difference calculation based on the lane identifier when the target vehicle passes through the current sensing device and the lane identifier when the multiple candidate vehicles disappear from the sensing range of the previous sensing device, to obtain the lane weights of the multiple candidate vehicles; Wherein, the trajectory behavior information corresponding to the target vehicle includes the time and lane identifier when the target vehicle passes through the sensing range of the current sensing device; the trajectory behavior information corresponding to the multiple candidate vehicles includes the time and lane identifier when the multiple candidate vehicles disappear from the sensing range of the previous sensing device.

2. The method according to claim 1, characterized in that, For any candidate vehicle, the time weight is used to characterize the probability that the candidate vehicle appears earlier than other candidate vehicles in the current sensing device; The lane weight is used to characterize the probability that the lanes where the multiple candidate vehicles are located are the same lane as the lane where the target vehicle is located.

3. The method according to claim 1, characterized in that, Based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when the multiple candidate vehicles disappear from the sensing range of the previous sensing device, performing a first difference calculation to obtain the time weights of the multiple candidate vehicles includes: Determine the cross-sensing device time of the multiple candidate vehicles based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when the multiple candidate vehicles disappear from the sensing range of the previous sensing device; Determine the time weights of the multiple candidate vehicles according to the cross-sensing device time of the multiple candidate vehicles.

4. The method according to claim 1, wherein The determining the feature weights of the multiple candidate vehicles includes: Based on the feature information of the target vehicle and the feature information of the multiple candidate vehicles, determine the similarity between the target vehicle and the multiple candidate vehicles, and determine the similarity corresponding to the multiple candidate vehicles as the feature weights; the feature information is extracted when the vehicle passes through the virtual detection line corresponding to the first sensing device.

5. The method according to claim 1, characterized in that, Before determining the matching vehicles that match the target vehicle among the multiple candidate vehicles based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively, it further includes: Determine that the difference between the two largest similarities among the similarities between the target vehicle and the multiple candidate vehicles is less than a preset threshold.

6. The method according to claim 5, characterized in that, The method further includes: In the case where the difference between the two largest similarities among the similarities between the target vehicle and the multiple candidate vehicles is greater than a preset threshold, determine the candidate vehicle with the highest similarity as the matching vehicle of the target vehicle.

7. The method according to any one of claims 1 to 6, characterized in that The determining the multiple candidate vehicles that match the target vehicle in terms of features among the multiple vehicles within the sensing range of the previous sensing device adjacent to the current sensing device includes: Determine the vehicles that meet the time constraint conditions and the space constraint conditions among the multiple vehicles, and determine the multiple candidate vehicles that match the target vehicle among the vehicles that meet the time constraint conditions and the space constraint conditions; Wherein, the space constraint conditions are determined based on the positions of the multiple vehicles, the distances between the target vehicle and the multiple vehicles, and the distance between the current sensing device and the previous sensing device; the time constraint conditions are determined based on the times when the target vehicle and the multiple vehicles pass through the sensing ranges of the current sensing device and the previous sensing device respectively.

8. The method according to claim 1, characterized in that, Before determining the matching vehicles that match the target vehicle among the multiple candidate vehicles based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively, the method further includes: Filter out the candidate vehicles that appear within the sensing range of the previous sensing device and do not disappear within the acquisition duration range.

9. The method according to claim 1, characterized in that The method further includes: In the case where there is a blind area between the sensing range of the previous sensing device and the sensing range of the current sensing device, based on the trajectory behavior information corresponding to the target vehicle and the matching vehicle respectively, determine the virtual driving trajectory of the virtual target vehicle corresponding to the target vehicle in the virtual environment, and the virtual driving trajectory includes the driving trajectory of the virtual target vehicle in the virtual blind area corresponding to the blind area.

10. A vehicle recognition device across sensing devices, characterized in that The device includes: A candidate vehicle determination module, configured to determine target vehicles within the sensing range of the current sensing device, and determine multiple candidate vehicles that match the characteristics of the target vehicle among multiple vehicles within the sensing range of the previous sensing device adjacent to the current sensing device; A matching vehicle determination module, configured to determine a matching vehicle that matches the target vehicle among the multiple candidate vehicles based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively; the trajectory behavior information includes attribute information of the position change behavior of the vehicle in the lane; wherein, the determining a matching vehicle that matches the target vehicle among the multiple candidate vehicles based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively includes: determining the time weight and lane weight of the multiple candidate vehicles based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively; determining a matching vehicle that matches the target vehicle among the multiple candidate vehicles based on the time weight and lane weight of the multiple candidate vehicles; the determining a matching vehicle that matches the target vehicle among the multiple candidate vehicles based on the time weight and lane weight of the multiple candidate vehicles includes: determining the feature weight of the multiple candidate vehicles; determining the matching vehicle of the target vehicle among the multiple candidate vehicles according to the time weight, the lane weight and the feature weight of the multiple candidate vehicles; The determining the time weight and lane weight of the multiple candidate vehicles based on the trajectory behavior information corresponding to the target vehicle and the multiple candidate vehicles respectively includes: Performing a first difference calculation based on the time when the target vehicle passes through the sensing range of the current sensing device and the time when the multiple candidate vehicles disappear from the sensing range of the previous sensing device to obtain the time weight of the multiple candidate vehicles; Performing a second difference calculation based on the lane identifier when the target vehicle passes through the current sensing device and the lane identifier when the multiple candidate vehicles disappear from the sensing range of the previous sensing device to obtain the lane weight of the multiple candidate vehicles; Wherein, the trajectory behavior information corresponding to the target vehicle includes the time and lane identifier when the target vehicle passes through the sensing range of the current sensing device; the trajectory behavior information corresponding to the multiple candidate vehicles includes the time and lane identifier when the multiple candidate vehicles disappear from the sensing range of the previous sensing device.

11. An electronic device, including a memory, a processor, and a computer program stored on the memory, where the processor implements the method according to any one of claims 1-9 when executing the computer program.

12. A computer-readable storage medium, in which a computer program is stored, and the computer program implements the method according to any one of claims 1-9 when executed by a processor.

Images