A vehicle recognition method, a vehicle recognition device, and a computer storage medium

Abstract

The present application provides a vehicle recognition method, a vehicle recognition device, and a computer storage medium. The vehicle recognition method includes: obtaining a vehicle trajectory to be recognized; extracting motion quantization features of the vehicle trajectory to be recognized; obtaining a confidence score of the motion quantization features; obtaining a probability score of the vehicle trajectory to be recognized based on the confidence score and a preset vehicle recognition model; and determining the vehicle type corresponding to the vehicle trajectory to be recognized according to the probability score. Through the above vehicle recognition method, a vehicle recognition model is designed, and the type of the vehicle is determined according to a preset probability threshold. The vehicle type can be quickly determined through real-time trajectory data, and the accuracy of vehicle recognition is improved.

Description

Technical Field

[0001] This application relates to the technical field of vehicle management, and particularly to a vehicle recognition method, a vehicle recognition device, and a computer storage medium. Background Art

[0002] Vehicle model recognition is an important part of the intelligent transportation management system and has a wide range of applications, such as traffic flow statistics, intelligent parking, and vehicle type detection. Therefore, a robust vehicle model recognition method is of great significance to the intelligent transportation system. Existing methods generally use technologies such as ultrasonic waves and magnetic induction coils. However, with the development of imaging technology and the widespread use of traffic management cameras, the image-based vehicle model recognition method has become the main trend.

[0003] However, the current vehicle model recognition methods all use manually designed general visual features, which are not specific to vehicle model recognition in the intelligent transportation scenario and cannot guarantee sufficient discriminative power. Therefore, they may fail in complex scenarios. Summary of the Invention

[0004] To solve the above technical problems, this application proposes a vehicle recognition method, a vehicle recognition device, and a computer storage medium.

[0005] To solve the above technical problems, this application proposes a vehicle recognition method, which includes:

[0006] Obtain the trajectory of the vehicle to be recognized;

[0007] Extract the motion quantization features of the trajectory of the vehicle to be recognized;

[0008] Obtain the confidence score of the motion quantization features;

[0009] Based on the confidence score and a preset vehicle recognition model, obtain the probability score of the trajectory of the vehicle to be recognized;

[0010] Determine the vehicle type corresponding to the trajectory of the vehicle to be recognized according to the probability score.

[0011] Wherein, after extracting the motion quantization features of the trajectory of the vehicle to be recognized, the vehicle recognition method further includes:

[0012] Obtain an attenuation function that decays according to the time distance from the current time;

[0013] Use the attenuation function to perform attenuation processing on the motion quantization features.

[0014] Wherein, the obtaining of the confidence score of the motion quantization features includes:

[0015] Obtain the logarithmic function value of the motion quantization feature;

[0016] Perform arctangent normalization on the logarithmic function value to obtain the confidence score.

[0017] Among them, the obtaining of the vehicle trajectory to be recognized includes:

[0018] Obtain the acquisition images of a number of acquisition points;

[0019] Identify the vehicle images to be recognized with the same license plate number in all acquisition images;

[0020] Obtain the sequence of vehicles to be recognized composed of all vehicle images to be recognized;

[0021] Use the sequence of vehicles to be recognized to generate the vehicle trajectory to be recognized.

[0022] Among them, the using of the sequence of vehicles to be recognized to generate the vehicle trajectory to be recognized includes:

[0023] Traverse the adjacent vehicle trajectory points in the sequence of vehicles to be recognized;

[0024] Eliminate one of the adjacent vehicle trajectory points with the acquisition time difference within a preset time period and the acquisition position difference less than the preset distance threshold to generate the vehicle trajectory to be recognized.

[0025] Among them, the using of the sequence of vehicles to be recognized to generate the vehicle trajectory to be recognized includes:

[0026] Obtain the sequence of targets to be recognized, where the sequence of targets to be recognized has the same license plate number information as the sequence of vehicles to be recognized;

[0027] Determine the first fusion weight based on the first acquisition quality score of the sequence of targets to be recognized;

[0028] Determine the second fusion weight based on the second acquisition quality score of the sequence of vehicles to be recognized

[0029] Use the first fusion weight and the second fusion weight to fuse the sequence of targets to be recognized and the sequence of vehicles to be recognized to generate the vehicle trajectory to be recognized.

[0030] Among them, the vehicle recognition method further includes:

[0031] Obtain the sequence of targets to be recognized, and / or the adjacent trajectory points in the sequence of targets to be recognized;

[0032] Eliminate one of the adjacent trajectory points with the speed change amount greater than the preset speed change threshold.

[0033] Among them, the step of removing one of the adjacent trajectory points with a speed change amount greater than a preset speed change threshold includes:

[0034] Removing one of the adjacent trajectory points with a speed change amount greater than a preset speed change threshold and a position angle change amount greater than a preset angle change threshold.

[0035] To solve the above technical problem, the present application also provides a vehicle recognition device, which includes a memory and a processor coupled to the memory; wherein, the memory is used to store program data, and the processor is used to execute the program data to implement the vehicle recognition method as described above.

[0036] To solve the above technical problem, the present application also provides a computer storage medium, which is used to store program data. When the program data is executed by a computer, it is used to implement the vehicle recognition method as described above.

[0037] Compared with the prior art, the beneficial effects of the present application are as follows: The vehicle recognition device obtains the trajectory of the vehicle to be recognized; extracts the motion quantization features of the trajectory of the vehicle to be recognized; obtains the confidence score of the motion quantization features; based on the confidence score and a preset vehicle recognition model, obtains the probability score of the trajectory of the vehicle to be recognized; and determines the vehicle type corresponding to the trajectory of the vehicle to be recognized according to the probability score. Through the above vehicle recognition method, a vehicle recognition model is designed, and the type of the vehicle is determined according to a preset probability threshold. The vehicle type can be quickly determined through real-time trajectory data, and the accuracy of vehicle recognition is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. Among them:

[0039] Figure 1 is a schematic flowchart of an embodiment of the vehicle recognition method provided by the present application;

[0040] Figure 2 is a schematic overall flowchart of the vehicle recognition method provided by the present application;

[0041] Figure 3 is a schematic flowchart of an embodiment of obtaining the trajectory of the vehicle to be recognized provided by the present application;

[0042] Figure 4 is Figure 1 a schematic detailed flowchart of step S11 of the vehicle recognition method shown;

[0043] Figure 5 It is a schematic flow chart of another embodiment for obtaining the vehicle trajectory to be recognized provided by this application;

[0044] Figure 6 It is a schematic diagram of an embodiment for vehicle trajectory fusion and completion provided by this application;

[0045] Figure 7 It is a schematic flow chart of an embodiment for the recognition method of the online car-hailing recognition model provided by this application;

[0046] Figure 8 It is a schematic diagram of the user interaction interface of the vehicle recognition system;

[0047] Figure 9 It is a schematic structural diagram of an embodiment of the vehicle recognition device provided by this application;

[0048] Figure 10 It is a schematic structural diagram of an embodiment of the computer storage medium provided by this application. Detailed implementation manners

[0049] Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this application.

[0050] The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and above-mentioned drawings of this application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of this application described here, for example, can be implemented in an order other than those illustrated or described here. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0051] This application provides a vehicle recognition system and method based on trajectory fusion in order to solve the problem of how to quickly recognize the type of vehicle and perform precise and efficient management and scheduling on it.

[0052] It should be noted that the vehicle type determined by the vehicle recognition method of the present application can be any vehicle type, including but not limited to: online car-hailing vehicles, ambulances, fire trucks, etc. In the subsequent embodiment introduction, online car-hailing vehicles are taken as an example for illustration.

[0053] The technologies involved in the vehicle recognition method of the present application are used for including but not limited to:

[0054] Online car-hailing vehicle: A fully network-reserved taxi, which is a new transportation mode that provides travel services by reserving platform taxis or private cars through an Internet platform.

[0055] Collection: A technology that uses cameras installed at specific positions to automatically capture and record image or video data at specific time points or time periods.

[0056] Trajectory: The path or route that a vehicle moves within a specific time period, including numbers and GPS (Global Positioning System) coordinate sequences.

[0057] RFM model: A commonly used analysis method for refined user operation. R (Recency, the time interval of the last consumption) represents the time interval of the last consumption, F (Frequency, consumption frequency) represents the consumption frequency, and M (Monetary, consumption amount) represents the consumption amount.

[0058] Long-tail effect: A phenomenon where data category imbalance causes a small number of categories to account for the majority of samples, while the majority of categories only have a small number of samples, presenting a long tail on the quantity distribution graph.

[0059] For details, please refer to Figure 1 and Figure 2 , Figure 1 is a schematic flowchart of an embodiment of the vehicle recognition method provided by the present application, Figure 2 is the overall schematic flowchart of the vehicle recognition method provided by the present application.

[0060] The vehicle recognition method of the present application is applied to a vehicle recognition device. Among them, the vehicle recognition device of the present application can be a server, or a terminal device, or a system in which the server and the terminal device cooperate with each other. Correspondingly, each part included in the vehicle recognition device, such as each unit, sub-unit, module, and sub-module, can be all set in the server, or all set in the terminal device, or separately set in the server and the terminal device.

[0061] Further, the above-mentioned server can be hardware or software. When the server is hardware, it can be implemented as a distributed server cluster composed of multiple servers or as a single server. When the server is software, it can be implemented as multiple software or software modules, such as software or software modules for providing a distributed server, or as a single software or software module, which is not specifically limited herein.

[0062] As Figure 1 shown, the specific steps are as follows:

[0063] Step S11: Obtain the vehicle trajectory to be recognized.

[0064] In the embodiment of the present application, the vehicle recognition device obtains real-time acquisition images from the cameras at each acquisition point, and uses image recognition technology to extract vehicle targets and other recognition targets that can assist in analyzing the vehicle trajectory in the acquisition images.

[0065] Among them, the vehicle recognition device can generate a vehicle trajectory to be recognized by using the acquisition image containing the target vehicle, for discriminating the vehicle type of the target vehicle.

[0066] Specifically, the vehicle recognition device obtains a vehicle trajectory data set, that is, obtains vehicle and vehicle target trajectory data collected at multiple acquisition points to construct a local vehicle data set. Among them, the vehicle target is other information on the vehicle, including but not limited to the marked style, unique identifier, and temporary identifier on the vehicle, etc. Among them, the temporary identifier can be an item inside the vehicle recognized through the window, such as a doll, etc.

[0067] For the technical solution of obtaining the vehicle trajectory to be recognized in the present application, please specifically refer to Figure 3 , Figure 3 which is a schematic flowchart of an embodiment of obtaining the vehicle trajectory to be recognized provided by the present application.

[0068] As Figure 3 shown, first, the vehicle recognition device generates a vehicle trajectory set and a vehicle target trajectory set. Since the types of acquisition points at different locations and sections are different, the vehicle data is mainly divided into the following two parts: generating the driving trajectory of the vehicle according to the license plate information collected at the acquisition point, and the attributes of the vehicle trajectory set should at least include but not be limited to the license plate number, acquisition point number, acquisition point coordinates, acquisition time, etc.; generating a vehicle target trajectory according to the vehicle target data collected at the acquisition point, and using target recognition technology to recognize the target attributes and supplement the corresponding vehicle information. Finally, the attributes of the vehicle target trajectory should at least include but not be limited to the target number, license plate number, acquisition point number, acquisition point coordinates, acquisition time, etc. The trajectory data set can be updated in real time according to the newly added vehicle trajectory.

[0069] It should be noted that the collected images in the vehicle target trajectory set do not necessarily need to identify the license plate number, and it is also possible to only identify the target number, that is, the vehicle target trajectory set is mainly created with the vehicle target as the trajectory core.

[0070] Furthermore, before vehicle recognition, in order to improve the efficiency of vehicle recognition, the vehicle recognition device can also first eliminate the hot data in the above vehicle trajectory set and / or vehicle target trajectory set.

[0071] Specifically, in some popular areas or areas where vehicle operation is congested, there may be a large number of collections for the same vehicle or the same vehicle target, resulting in data redundancy. Therefore, the vehicle trajectory set and / or the vehicle target trajectory set need to remove continuous collections within a short period of time. Specifically, the collections of the same vehicle or the same vehicle target can be sorted in descending order of time, and the time difference and distance change between the collections are calculated. For the trajectory sequence data of the same set, only the trajectory sequence data with a change in the collection point position greater than the preset distance threshold within the adjacent preset time is retained. For example, the vehicle recognition device can only retain the collection data with a GPS change greater than 100m within adjacent 5 minutes.

[0072] The vehicle recognition device can extract the vehicle trajectory set and the vehicle target trajectory set at the same time. Given that the collection points at the current intersection have the function of collecting vehicles and vehicle targets, and can form a set based on the vehicle target to obtain real-name information, the vehicle recognition device can choose to fuse the trajectories of the vehicle and the vehicle target, making up for the possible missing of vehicle trajectories, so that the vehicle trajectories are more continuous and the feature information is more abundant.

[0073] For details, please refer to Figure 4 and Figure 5 , Figure 4 is Figure 1 the specific process schematic diagram of step S11 of the vehicle recognition method shown, Figure 5 is the process schematic diagram of another embodiment for obtaining the trajectory of the vehicle to be recognized provided by this application.

[0074] As Figure 4 shown, the specific steps are as follows:

[0075] Step S111: Obtain the target sequence to be recognized, where the target sequence to be recognized has the same license plate number information as the vehicle sequence to be recognized.

[0076] In the embodiment of this application, the vehicle recognition device obtains the target sequence to be recognized according to the vehicle target trajectory set, and obtains the vehicle sequence to be recognized according to the vehicle trajectory set. The above two sequences are associated through the same license plate number.

[0077] Furthermore, as Figure 5As shown, before fusing and complementing the vehicle trajectory, in order to improve the accuracy of the vehicle trajectory, the vehicle recognition device may also remove the trajectory abnormal data in the to-be-recognized target sequence and / or the to-be-recognized vehicle sequence.

[0078] Due to the influence of factors such as the acquisition angle, light, and others, there may be abnormal data in the vehicle and vehicle target trajectories, which affects subsequent model analysis. Therefore, it is necessary to clean the obtained vehicle trajectory dataset and remove the abnormal trajectory data.

[0079] Taking the to-be-recognized vehicle sequence as an example, the vehicle recognition device sorts the trajectory data of the to-be-recognized vehicle sequence according to the time sequence, and calculates the speed change and angle change between adjacent trajectory points. If within a certain time interval, both the speed change and the angle change exceed the preset threshold, then there is a situation of abnormal trajectory for this adjacent trajectory point, and the data of this adjacent trajectory point is removed.

[0080] Specifically, the speed abnormality calculation formula is as follows:

[0081]

[0082] Where, is the speed change amount between adjacent trajectory points, is the speed change threshold.

[0083] Specifically, the angle abnormality calculation formula is as follows:

[0084]

[0085] Where, is the angle change amount between adjacent trajectory points, is the angle change threshold.

[0086] In summary, the vehicle recognition device can choose to remove the speed abnormal trajectory points and / or angle abnormal trajectory points calculated above to update the to-be-recognized vehicle sequence.

[0087] Step S112: Determine the first fusion weight based on the first acquisition quality score of the to-be-recognized target sequence.

[0088] Step S113: Determine the second fusion weight based on the second acquisition quality score of the to-be-recognized vehicle sequence.

[0089] In the embodiments of the present application, the vehicle recognition device uses the to-be-recognized target sequence to perform vehicle trajectory fusion and complementation on the to-be-recognized vehicle sequence. Specifically, due to the actual situation that the acquisition points are not comprehensively covered, there are easily situations where the trajectory data is blank for a long time or the trajectory jumps, resulting in the problem of low subsequent model recognition rate. Therefore, it is necessary to fuse and complement the overall trajectory by combining the vehicle acquisition data and the vehicle target acquisition data.

[0090] Specifically, as Figure 6 shown, Figure 6 it is a schematic diagram of an embodiment of vehicle trajectory fusion and completion provided by this application. As Figure 6 shown, if the vehicle recognition device identifies abnormal trajectory points in the vehicle trajectory through step S111, the abnormal trajectory points are first removed, and then the vehicle trajectory and the vehicle target trajectory are fused to obtain the final fused trajectory, that is, the vehicle trajectory to be recognized shown in step S11.

[0091] Specifically, the vehicle recognition device obtains the vehicle trajectory and the vehicle target trajectory under the same license plate. If the two trajectories use different coordinate systems, coordinate space conversion is required. The vehicle recognition device calculates the quality score for each image to evaluate the quality of the current acquisition. By calculating the average quality score of all the acquired images in the two types of trajectories and and normalizing it, it is used as the calculation weight for subsequent trajectory fusion.

[0092] Among them, the quality scores of the acquired images of the vehicle trajectory and the vehicle target trajectory can be evaluated by the image clarity, the integrity of the vehicle or the vehicle target.

[0093] Therefore, the vehicle trajectory and the vehicle target trajectory are respectively expressed as:

[0094]

[0095] Step S114: Using the first fusion weight and the second fusion weight, fuse the target sequence to be recognized and the vehicle sequence to be recognized to generate the vehicle trajectory to be recognized.

[0096] Furthermore, the vehicle recognition device obtains the overall trajectory time range, that is, the start time and the end time of the vehicle trajectory to be recognized:

[0097]

[0098] The vehicle recognition device performs interpolation fusion on the trajectory according to the calculation weight:

[0099]

[0100] After the above processing, two or more trajectories corresponding to the license plate can be effectively fused into one trajectory to obtain complete vehicle running trajectory information.

[0101] In this regard, by repeatedly performing the above processing process on the vehicle trajectory data of different dates, the vehicle recognition device can obtain the trajectory data sequence of each vehicle for multiple days.

[0102] Step S12: Extract the motion quantization features of the vehicle trajectory to be recognized.

[0103] In the embodiment of the present application, please continue to refer to Figure 5 , the vehicle recognition device constructs motion quantization features such as hot spot data and boarding / alighting points based on the vehicle trajectory to be recognized after fusion and completion. Among them, the motion quantization feature refers to the feature that measures and evaluates the law of vehicle motion and vehicle motion points, etc. through specific numerical values and indicators during the vehicle motion process, and is used to characterize the motion information of the vehicle.

[0104] Specifically, since the online car-hailing has a fixed driving pattern, which is mainly manifested as frequently traveling back and forth between hot spot areas and the boarding / alighting points often concentrated in hot spot areas, the vehicle recognition device can obtain the hot spot areas where the vehicle travels according to this feature through the trajectory data.

[0105] First, the vehicle recognition device identifies the parking points through a time window-based method and obtains all the parking point data of the vehicle in one day. Specifically, the vehicle recognition device sets a time window, and if the vehicle stays at a certain position for more than the threshold, then this position is considered as a parking point.

[0106] Then, the vehicle recognition device uses a clustering algorithm to identify the hot spot areas in the trajectory and counts the number of parking times of the vehicle in the hot spot areas. Specifically, such as the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm.

[0107] Finally, the vehicle recognition device counts the driving frequency and total driving distance of the vehicle in different time periods (including morning and evening rush hours) to estimate the number of trips of the vehicle. Specifically, the vehicle recognition device can divide a day into 48 time intervals at 30-minute intervals and count the number of times the same vehicle is collected in each time interval as the driving frequency of the vehicle.

[0108] Please continue to refer to Figure 7 , Figure 7 is a schematic flowchart of an embodiment of the recognition method of the online car-hailing recognition model provided by the present application.

[0109] As Figure 7 shown, the vehicle recognition device quantifies the vehicle trajectory data, obtains the online car-hailing recognition model, and recognizes the vehicle according to the confidence level. Specifically, the vehicle recognition device quantifies the vehicle trajectory data, and through the RFM model analysis, the following three-dimensional features can be obtained: Rencency feature, Frequency feature, Monetary feature.

[0110] Specifically, the Rencency feature represents the number of parking points in the hot zone in the most recent date of the vehicle trajectory set. The larger the number, the more active it indicates. The Frequency feature represents the number of days when trajectory collection occurs within a certain period of the vehicle trajectory set. The higher the number, the more active it indicates. The Monetary feature represents the driving frequency and total mileage within a certain period of the vehicle trajectory set. The higher the driving frequency and total mileage, the more active it indicates.

[0111] The features of each of the above dimensions represent the cumulative behavior over a past period of time. To make the behavior closer to the present have a greater impact on the activity score, the vehicle recognition device can also multiply the data of each day by a weight and then sum them. This weight is a function that decays as the time from the present increases. The specific decay formula is as follows:

[0112]

[0113] Among them, represents the decay function, represents the feature of any one of the above dimensions.

[0114] Step S13: Obtain the confidence score of the motion quantization feature.

[0115] In the embodiment of the present application, to increase the discrimination of data and avoid the data long-tail effect, the vehicle recognition device processes the features extracted in step S12 through a function. Subsequently, the vehicle recognition device performs arctangent normalization processing on the feature data to unify the data of different dimensions to the same dimension, facilitating the subsequent calculation of the trajectory judgment confidence score.

[0116] The specific calculation formula is as follows:

[0117]

[0118] Among them, is the confidence score of each dimension feature.

[0119] Furthermore, since the present application can evaluate the motion quantization features of different dimensions of the vehicle trajectory to be recognized, therefore, the vehicle recognition device needs to calculate the confidence score by weighting, assign a weight to the score of each dimension feature, and sum to obtain the final confidence score of the target vehicle as a network car:

[0120]

[0121] Among them, is the weight, which can be obtained through training with labeled data or given manually.

[0122] Step S14: Based on the confidence score and the preset vehicle recognition model, obtain the probability score of the vehicle trajectory to be recognized.

[0123] In the embodiment of the present application, based on the vehicle confidence score obtained in step S13, the vehicle recognition device can judge the vehicle trajectory through a binary classification / or multi-classification analysis model.

[0124] Specifically, the vehicle recognition device can generate a vehicle judgment model through a logistic regression analysis method, map the output of the linear combination to between 0 and 1, and obtain whether the vehicle is a online car-hailing vehicle:

[0125]

[0126] Among them, represents the weight of this category on the input features and can be learned through training data; represents the bias term, which can be regarded as the constant term of the model and is used to adjust the baseline probability of each category; represents the confidence score, which contains all the features of the vehicle trajectory to be recognized, and each feature affects the result of the model through the above-mentioned weight; represents the total number of categories, which is 2 in the binary classification problem; represents the probability that the vehicle is an online car-hailing vehicle.

[0127] Step S15: Determine the vehicle type corresponding to the vehicle trajectory to be recognized according to the probability score.

[0128] In the embodiment of the present application, the vehicle recognition device determines whether the vehicle is an online car-hailing vehicle based on the probability threshold set for online car-hailing vehicles.

[0129] In other embodiments, the vehicle recognition device can also set corresponding probability thresholds for other vehicle types, and the determination process is as described in the above steps S11 to S15, which will not be elaborated here.

[0130] Furthermore, the present application also provides a user interaction interface. For details, please refer to Figure 8 , Figure 8 is a schematic diagram of the user interaction interface of the vehicle recognition system. As Figure 8 shown, the user selects any vehicle in the vehicle details through this interaction interface, and the interaction interface will display the trajectory situation and the specific trajectory details in a graphical manner, specifically including the captured images of each trajectory point, etc.

[0131] In this application, a vehicle recognition device acquires the trajectory of a vehicle to be recognized; extracts the motion quantization features of the trajectory of the vehicle to be recognized; obtains the confidence score of the motion quantization features; based on the confidence score and a preset vehicle recognition model, obtains the probability score of the trajectory of the vehicle to be recognized; and determines the vehicle type corresponding to the trajectory of the vehicle to be recognized according to the probability score. Through the above vehicle recognition method, a vehicle recognition model is designed, and the type of the vehicle is determined according to a preset probability threshold. The vehicle type can be quickly determined through real-time trajectory data, and the accuracy of vehicle recognition is improved.

[0132] The vehicle recognition method of this application fills in the gaps in trajectory data through the combination of multiple trajectories, making the data utilization more effective and the recognition more accurate.

[0133] The vehicle recognition method of this application uses real-time acquisition devices such as video acquisition points that are widely laid to achieve full coverage of the area, avoid the lack of vehicle trajectories, and achieve real-time and accurate entity management.

[0134] The vehicle recognition method of this application designs a network car recognition model, determines whether a vehicle is a network car according to a set probability threshold; deeply integrates trajectory data, data mining technology and traffic services to effectively identify network cars and improve traffic management efficiency.

[0135] Those skilled in the art can understand that in the above method of the specific embodiment, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined according to its function and possible internal logic.

[0136] To implement the above vehicle recognition method, this application also proposes a vehicle recognition device. For details, please refer to Figure 9 , Figure 9 is a schematic structural diagram of an embodiment of the vehicle recognition device provided by this application.

[0137] The vehicle recognition device 400 of this embodiment includes a processor 41, a memory 42, an input / output device 43, and a bus 44.

[0138] The processor 41, the memory 42, and the input / output device 43 are respectively connected to the bus 44. Program data is stored in the memory 42, and the processor 41 is used to execute the program data to implement the vehicle recognition method described in the above embodiment.

[0139] In an embodiment of the present application, the processor 41 may also be referred to as a CPU (Central Processing Unit). The processor 41 may be an integrated circuit chip with signal processing capabilities. The processor 41 may also be a general-purpose processor, a digital signal processor (DSP, Digital Signal Process), an application-specific integrated circuit (ASIC, Application Specific Integrated Circuit), a field-programmable gate array (FPGA, Field Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The general-purpose processor may be a microprocessor or the processor 41 may also be any conventional processor, etc.

[0140] The present application also provides a computer storage medium. Please continue to refer to Figure 10 , Figure 10 FIG. is a schematic structural diagram of an embodiment of the computer storage medium provided by the present application. A computer program 61 is stored in the computer storage medium 600. When the computer program 61 is executed by a processor, it is used to implement the vehicle recognition method in the above embodiment.

[0141] When the embodiments of the present application are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0142] The above are only the embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present application.

Claims

1. A vehicle identification method, characterized in that: The vehicle identification method comprises: Obtain the trajectory of the vehicle to be identified; Extracting the motion quantitative features of the to-be-identified vehicle trajectory; wherein the motion quantitative features refer to measuring and evaluating the law of vehicle motion and the features of vehicle motion points through specific values ​​and indicators during the vehicle motion process, which are used to characterize the vehicle motion information; Obtaining a confidence score of the motion quantization feature; Based on the confidence score and a preset vehicle recognition model, obtaining a probability score of the to-be-recognized vehicle trajectory; Determining the vehicle type corresponding to the to-be-identified vehicle trajectory according to the probability score; The obtaining of the confidence score of the motion quantization feature comprises: Obtaining a logarithmic function value of the motion quantization feature; The logarithmic function value is normalized by arctangent to obtain the confidence score.

2. The vehicle identification method according to claim 1, characterized in that: After extracting the motion quantization feature of the to-be-identified vehicle trajectory, the vehicle identification method further includes: Get a decay function that decays as the time distance from the current time increases; The motion quantization feature is attenuated using the attenuation function.

3. The vehicle identification method according to claim 1, characterized in that: The step of obtaining the trajectory of the vehicle to be identified includes: Acquire collected images of several collection points; Identify the vehicle image to be identified with the same license plate number in all collected images; Obtain a sequence of vehicles to be identified consisting of all images of vehicles to be identified; The to-be-identified vehicle trajectory is generated by using the to-be-identified vehicle sequence.

4. The vehicle identification method according to claim 3, characterized in that: The step of generating the to-be-identified vehicle trajectory by using the to-be-identified vehicle sequence comprises: Traversing adjacent vehicle trajectory points in the sequence of vehicles to be identified; One of the adjacent vehicle trajectory points whose acquisition time difference is within a preset time length and whose acquisition position difference is less than a preset distance threshold is eliminated to generate the vehicle trajectory to be identified.

5. The vehicle identification method according to claim 3, characterized in that: The step of generating the to-be-identified vehicle trajectory by using the to-be-identified vehicle sequence comprises: Acquire a target sequence to be identified, wherein the target sequence to be identified and the vehicle sequence to be identified have the same license plate number information; Determining a first fusion weight based on a first acquisition quality score of the target sequence to be identified; Determine a second fusion weight based on the second acquisition quality score of the vehicle sequence to be identified The first fusion weight and the second fusion weight are used to fuse the to-be-identified target sequence and the to-be-identified vehicle sequence to generate the to-be-identified vehicle trajectory.

6. The vehicle identification method according to claim 5, characterized in that: The vehicle identification method further includes: Acquire the target sequence to be identified, and / or adjacent trajectory points in the target sequence to be identified; One of the adjacent trajectory points whose speed change is greater than a preset speed change threshold is removed.

7. The vehicle identification method according to claim 6, characterized in that: The step of removing one of the adjacent trajectory points whose speed change is greater than a preset speed change threshold comprises: One of the adjacent track points whose speed change is greater than a preset speed change threshold and whose position angle change is greater than a preset angle change threshold is removed.

8. A vehicle identification device, characterized in that: The vehicle identification device includes a memory and a processor coupled to the memory; The memory is used to store program data, and the processor is used to execute the program data to implement the vehicle identification method as described in any one of claims 1 to 7.

9. A computer storage medium, characterized in that The computer storage medium is used to store program data, and when the program data is executed by a computer, it is used to implement the vehicle identification method according to any one of claims 1 to 7.

Images