Vehicle tracking method, electronic equipment and storage medium

Abstract

The embodiment of the invention relates to the technical field of vehicle tracking, and particularly relates to a vehicle tracking method, electronic equipment and a storage medium, and the method comprises the steps: obtaining a plurality of original images, extracting the driving data and feature data of a vehicle based on the plurality of original images, and obtaining a vehicle tracking result based on the driving data and the feature data; calculating the similarity score of the candidate vehicle in the candidate original image, determining that the target vehicle, the target vehicle and the reference vehicle are the same vehicle based on the similarity score, obtaining the initial positions of the target vehicle and the reference vehicle from the driving data, and obtaining the target position based on the initial positions of the target vehicle and the reference vehicle. According to the embodiment of the invention, the driving data and the feature data of the vehicle are extracted from the plurality of original images, the same vehicle is determined based on the driving data and the feature data, more accurate target position information of the vehicle is obtained according to the initial position information of the same vehicle, the vehicle can be continuously and accurately tracked, and the tracking efficiency is improved. And the vehicle tracking accuracy is improved.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of vehicle tracking, and in particular to a vehicle tracking method, an electronic device and a storage medium. BACKGROUND

[0002] With the development of social economy and the progress of science and technology, the urbanization process in China is accelerating, and the challenges faced by urban traffic are becoming increasingly severe. The current intelligent traffic monitoring method mainly relies on laying cameras and sensors on the road, and with the development of image recognition technology, vehicle target detection and tracking technology based on video images has achieved certain results, but its limitations are increasingly highlighted in the face of increasing traffic flow and increasingly complex traffic conditions.

[0003] At present, for the detection and tracking problem of moving vehicles between multiple cameras, only simple scenarios with few moving vehicles and slow speed are considered, without involving complex feature selection problems of vehicle tracking, differences in resolution, lighting, viewing angle and background environment of multiple cameras, resulting in inability to accurately and continuously track vehicles and difficulty in obtaining accurate vehicle tracking results. SUMMARY

[0004] Therefore, an embodiment of the present application aims to provide a vehicle tracking method, an electronic device and a storage medium to solve the technical problem that the prior art cannot accurately track vehicles.

[0005] To solve the above technical problems, an embodiment of the present application provides the following technical solutions: In a first aspect, an embodiment of the present application provides a vehicle tracking method, comprising: obtaining vehicle image data, the vehicle image data comprising a plurality of original images, the plurality of original images being images obtained by a plurality of cameras arranged at different spatial positions at the same time; extracting driving data and feature data of a vehicle based on the plurality of original images, the driving data being data for representing a driving state of the vehicle, and the feature data being data for representing an appearance feature of the vehicle; calculating a similarity score of a candidate vehicle in a candidate original image based on the driving data and the feature data, the similarity score representing a similarity between the candidate vehicle and a reference vehicle in a reference original image, the candidate original image and the reference original image being any two original images in the plurality of original images, the candidate vehicle being any vehicle in the candidate original image, and the reference vehicle being any vehicle in the reference original image; determine a target vehicle based on the similarity score, the target vehicle being a candidate vehicle corresponding to the similarity score meeting a preset score requirement, the target vehicle and the reference vehicle being the same vehicle; obtain initial positions of the target vehicle and the reference vehicle from the driving data, and obtain a target position based on the initial positions of the target vehicle and the reference vehicle, the target position representing positions of the target vehicle and the reference vehicle.

[0006] In some embodiments, the similarity score of the candidate vehicle in the candidate original image is calculated based on the driving data and the feature data, including: obtain a confidence score of the candidate vehicle based on the feature data of the candidate vehicle; obtain a matching degree score of the candidate vehicle and the reference vehicle based on the feature data of the candidate vehicle and the reference vehicle; obtain a motion trajectory and a stability score of the candidate vehicle based on the driving data of the candidate vehicle and the reference vehicle, the stability score representing stability of the motion trajectory; obtain the similarity score of the candidate vehicle based on the confidence score, the matching degree score and the stability score.

[0007] In some embodiments, the feature data includes size, shape and color, and the confidence score of the candidate vehicle is obtained based on the feature data of the candidate vehicle, including: compare the size, shape and color of the candidate vehicle with standard size, standard shape and standard color of a preset vehicle model respectively to obtain a first comparison result, the first comparison result being the confidence score, the confidence score representing credibility of the candidate vehicle being a real vehicle.

[0008] In some embodiments, the feature data further includes a logo and a license plate, and the matching degree score of the candidate vehicle and the reference vehicle is obtained based on the feature data of the candidate vehicle and the reference vehicle, including: compare the size, shape, color, logo and license plate of the candidate vehicle with reference size, reference shape, reference color, reference logo and reference license plate of the reference vehicle respectively to obtain a second comparison result, the second comparison result being the matching degree score, wherein the matching degree score represents matching degree of the candidate vehicle and the reference vehicle.

[0009] In some embodiments, the driving data includes initial position, speed, acceleration and driving direction, and the motion trajectory and the stability score of the candidate vehicle are obtained based on the driving data of the candidate vehicle and the reference vehicle, including: calculate a motion trajectory of the candidate vehicle and a stability score of the motion trajectory based on the initial position, the speed, the acceleration and the driving direction of the candidate vehicle, the reference initial position, the reference speed, the reference acceleration and the reference driving direction of the reference vehicle.

[0010] In some embodiments, the obtaining the similarity score of the candidate vehicle based on the confidence score, the matching degree score and the stability score comprises: weighting and fusing the confidence score, the matching degree score and the stability score to obtain the similarity score of the candidate vehicle.

[0011] In some embodiments, the determining the target vehicle based on the similarity score comprises: selecting a similarity score greater than a preset score threshold and being a maximum value among a plurality of the similarity scores as a target similarity score; determining a candidate vehicle corresponding to the target similarity score as the target vehicle.

[0012] In some embodiments, the original image in which the target vehicle is located is a target original image, each of the camera devices corresponds to one of the original images, and the obtaining the target position based on the initial positions of the target vehicle and the reference vehicle comprises: constructing a weighting fusion formula based on a target initial position and a reference initial position, the target initial position and the reference initial position being the initial positions of the target vehicle and the reference vehicle respectively, the weighting fusion formula comprising the target initial position, the reference initial position, a weight coefficient of the target initial position and a weight coefficient of the reference initial position; obtaining a target extrinsic calibration precision, a target resolution and a target picture position of the target vehicle in the target original image of a target camera device, the target camera device being a camera device corresponding to the target original image; obtaining a reference extrinsic calibration precision, a reference resolution and a reference picture position of the reference vehicle in a reference original image of a reference camera device, the reference camera device being a camera device corresponding to the reference original image; calculating the weight coefficient of the target initial position based on the target extrinsic calibration precision, the target resolution and the target picture position; calculating the weight coefficient of the reference initial position based on the reference extrinsic calibration precision, the reference resolution and the reference picture position; substituting the target initial position and the reference initial position into the weighting fusion formula to obtain the target position.

[0013] In a second aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory connected to the processor in communication; The memory stores computer program instructions executable by the processor, which, when executed by the processor, causes the electronic device to perform any one of the vehicle tracking methods of the first aspect.

[0014] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, which stores computer program instructions executable by a processor, which, when executed by the processor, causes the computer to perform any one of the vehicle tracking methods of the first aspect.

[0015] The embodiments of the present application have the following beneficial effects: Different from the prior art, the vehicle tracking method provided by the embodiments of the present application comprises: obtaining vehicle image data, the vehicle image data comprising a plurality of original images, the plurality of original images being images obtained by a plurality of camera devices arranged at different spatial positions at the same time, extracting driving data and feature data of the vehicle in the original images, the driving data being data for representing a driving state of the vehicle, and the feature data being data for representing an appearance feature of the vehicle, calculating a similarity score of a candidate vehicle in a candidate original image based on the driving data and the feature data, the similarity score representing a similarity between the candidate vehicle and a reference vehicle in a reference original image, the candidate original image and the reference original image being any two original images in the plurality of original images, the candidate vehicle being any one vehicle in the candidate original image, and the reference vehicle being any one vehicle in the reference original image, determining a target vehicle based on the similarity score, the target vehicle being a candidate vehicle corresponding to a similarity score meeting a preset score requirement, the target vehicle and the reference vehicle being the same vehicle, obtaining an initial position of the target vehicle and the reference vehicle from the driving data, and obtaining a target position based on the initial position of the target vehicle and the reference vehicle, the target position representing a position of the target vehicle and the reference vehicle.

[0016] The embodiments of the present application extract the driving data and the feature data of the vehicle from the plurality of original images obtained by the plurality of camera devices at the same time, determine the same vehicle in the plurality of original images based on the driving data and the feature data, and obtain more accurate position information of the vehicle according to the initial position information of the same vehicle in the plurality of original images, so that the vehicle can be continuously and accurately tracked, and the accuracy of vehicle tracking is improved. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the prior art or the embodiments. Obviously, the following described drawings only show some of the embodiments of the present application, and should not be regarded as a limitation to the protection scope. For those skilled in the art, other related drawings can also be obtained from these drawings without creative labor.

[0018] Figure 1 is a schematic diagram of an application scenario of a vehicle tracking method provided by some embodiments of the present application; Figure 2a is a first original image captured by any one of the camera devices in some embodiments of the present application; Figure 2b is a second original image captured by another camera device in some embodiments of the present application; Figure 3 is a schematic diagram of the structure of an electronic device provided by some embodiments of the present application; Figure 4 is a schematic diagram of a flow of a vehicle tracking method provided by some embodiments of the present application; Figure 5 is Figure 2b is a schematic diagram of the second original image divided into a plurality of sub-picture images; Figure 6 is Figure 2a is a schematic diagram of the first original image divided into a plurality of sub-picture images. DETAILED DESCRIPTION

[0019] In order to make the objects and advantages of the embodiments of the present application more easily understood, the technical solutions of the embodiments of the present application will be described clearly and completely below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. The following detailed description of the embodiments of the present application in the drawings is not intended to limit the scope of the present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0020] It should be noted that, if no conflict is constituted, the various technical features involved in the embodiments of the present invention described below can be combined with each other and are all within the scope of protection of the present invention. In addition, although the functional modules are divided in the device or structural diagram and the logical order is shown in the flow chart, in some cases, the steps shown or described can be performed in a different module division than in the device or in an order different from that in the flow chart. In addition, the "first", "second", "third" and other similar expressions used herein do not limit the data and execution order, but are only for the purpose of convenience of explanation and to distinguish between the same items or similar items with basically the same functions and effects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features.

[0021] Unless otherwise defined, the technical and scientific terms used in this specification have the same meanings as those commonly understood by those skilled in the art within the technical field of the present invention. The terms used in this specification are intended solely to describe specific embodiments and are not intended to limit the present invention. It should be understood that the term "and / or" as used in this specification includes any and all combinations of one or more of the listed items.

[0022] See also Figure 1 , Figure 1 The following schematically illustrates an application scenario of a vehicle tracking method provided by some embodiments of the present invention.

[0023] like Figure 1 As shown, the application scenario includes an electronic device 100 and a camera device 200. The electronic device 100 is connected to the camera device 200 via a network. Examples of the network include but are not limited to the Internet, corporate intranet, local area network, mobile communication network and combinations thereof.

[0024] Exemplarily, there are at least two cameras 200, which are arranged at different spatial locations. The cameras 200 are used to capture vehicle image data. The vehicle image data includes multiple original images captured simultaneously by the at least two cameras 200. It is understood that the original images may or may not include the vehicle.

[0025] After capturing and acquiring vehicle image data, the camera device 200 transmits the vehicle image data to the electronic device 100 via a network. Based on the multiple original images, the electronic device 100 extracts the vehicle's driving data and feature data. Specifically, the electronic device 100 extracts the vehicle's driving data and feature data from each original image. It should be understood that driving data is data used to characterize the vehicle's driving state, such as its speed, acceleration, direction of travel, and initial position. Feature data is data used to characterize the vehicle's appearance, such as its shape, size, and color.

[0026] Specifically, after extracting the driving data and feature data of the vehicle, the electronic device 100 calculates the similarity score of the candidate vehicle in the candidate original image based on the driving data and feature data of the vehicle, the similarity score representing the similarity between the candidate vehicle and the reference vehicle in the reference original image, the candidate original image and the reference original image being any two original images in the plurality of original images, the candidate vehicle being any vehicle in the candidate original image, and the reference vehicle being any vehicle in the reference original image. That is, any two original images in the plurality of original images are selected as the candidate original image and the reference original image, respectively, any vehicle in the reference original image is selected as the reference vehicle, any vehicle in the candidate original image is selected as the candidate vehicle, and the similarity score of the candidate vehicle is calculated based on the driving data and feature data of the reference vehicle and the driving data and feature data of the candidate vehicle.

[0027] After calculating the similarity scores of all candidate vehicles in the candidate original image, the similarity scores meeting the preset score requirement are screened out, and the candidate vehicle corresponding to the similarity score meeting the preset score requirement is taken as the target vehicle. It can be understood that the target vehicle and the reference vehicle are the same vehicle in the candidate original image and the reference original image. In the embodiment of the present application, the preset score requirement is that the similarity score is greater than the preset similarity score threshold and the similarity score is the maximum value among the similarity scores of all candidate vehicles.

[0028] For example, the initial position of the target vehicle is extracted from the driving data of the target vehicle, the initial position of the reference vehicle is extracted from the driving data of the reference vehicle, and the target position is obtained according to the initial position of the target vehicle and the initial position of the reference vehicle, the target position being more accurate position information of the target vehicle and the reference vehicle. In this way, continuous and accurate tracking of the vehicle is achieved, and the accuracy of vehicle tracking is improved.

[0029] For example, please refer to Figure 2a and Figure 2b , Figure 2a schematically shows the first original image 201 obtained by photographing any one camera, Figure 2bThe second original image 202 obtained by another camera is shown schematically, the first original image 201 includes the vehicle 11, the vehicle 12, the vehicle 13, the vehicle 14 and the tree 15, and the second original image 202 includes the vehicle 12, the vehicle 14, the tree 15, the tree 16, the vehicle 17 and the vehicle 18. The first original image 201 is taken as a reference original image, the second original image 202 is taken as a candidate original image, the vehicle 11 in the first original image 201 is determined as a reference vehicle, and the vehicle 12, the vehicle 14, the tree 15, the tree 16, the vehicle 17 and the vehicle 18 in the second original image 202 are determined as candidate vehicles respectively, then the driving data and the feature data of each candidate vehicle are compared with the driving data and the feature data of the reference vehicle 11, the similarity scores of each candidate vehicle are calculated, and whether each candidate vehicle is the same vehicle as the reference vehicle 11 is determined according to the similarity scores.

[0030] In the embodiment of the present application, the similarity score threshold is set to 0.90, and the similarity scores of the candidate vehicle 12, the candidate vehicle 14, the candidate vehicle 15, the candidate vehicle 16, the candidate vehicle 17 and the candidate vehicle 18 are 0.65, 0.65, 0.10, 0.10, 0.60 and 0.50 respectively. Obviously, the similarity scores of all the candidate vehicles are less than the similarity score threshold 0.90, so all the vehicles in the candidate original image (i.e. the second original image 202) and the reference vehicle 11 are not the same vehicle, that is, there is no vehicle highly similar to the reference vehicle 11 in the second original image 202.

[0031] For example, other vehicles in the first original image 201 are then determined as reference vehicles, for example, the vehicle 12 is determined as a reference vehicle, and the similarity scores of each candidate vehicle are calculated according to the driving data and the feature data of each candidate vehicle in the second original image 202 and the driving data and the feature data of the reference vehicle 12, and whether each candidate vehicle is the same vehicle as the reference vehicle 12 is determined according to the similarity scores. Among them, the similarity scores of the candidate vehicle 12, the candidate vehicle 14, the candidate vehicle 15, the candidate vehicle 16, the candidate vehicle 17 and the candidate vehicle 18 are 0.95, 0.65, 0.10, 0.10, 0.60 and 0.50 respectively. Obviously, the similarity score of the candidate vehicle 12 is greater than the similarity score threshold 0.90 and is the maximum among the similarity scores of all the candidate vehicles, so the candidate vehicle 12 in the candidate original image (i.e. the second original image 202) and the reference vehicle 12 are the same vehicle, and the candidate vehicle 12 is the target vehicle.

[0032] Finally, the initial position of the target vehicle 12 (i.e., the candidate vehicle 12) and the initial position of the reference vehicle 12 are extracted, and according to the initial position of the target vehicle 12 and the initial position of the reference vehicle 12, a target position is obtained, which represents more accurate position information of the target vehicle 12 and the reference vehicle 12.

[0033] It should be understood that, in Figure 1 the application scenarios shown, the electronic device 100 is a notebook computer, but it does not cause any limitation to any situation of the structure, type, and number of the electronic device in other application scenarios or embodiments. For example, in some other application scenarios or embodiments, the electronic device can also be a tablet computer, a desktop computer, or any other suitable type of device, or the electronic device can also be a server, such as a server deployed in the cloud. In addition, the electronic device in some other application scenarios or embodiments can also have more or fewer components than the notebook computer shown in Figure 1 , or have a different configuration from the notebook computer shown in Figure 1 .

[0034] To facilitate understanding of the vehicle tracking method provided by the embodiments of the present application, first, the electronic device provided by the embodiments of the present application is described in detail.

[0035] Please refer to Figure 3 , Figure 3 schematically shows the structure of the electronic device provided by some embodiments of the present application.

[0036] As shown in Figure 3 , the electronic device 100 includes at least one processor 110 and a memory 120 connected in communication, Figure 3 , a processor is taken as an example. Among them, each component in the electronic device 100 is coupled together through the bus system 130, and the bus system 130 is used to realize the connection and communication between each component. It is easy to understand that the bus system 130 can include a power bus, a control bus, and a state signal bus in addition to a data bus. However, in order to clearly illustrate and concisely, all kinds of buses are marked as the bus system 130 in Figure 3 . It can be understood that, Figure 3 the structure shown in the embodiments is only schematic, and it does not cause any limitation to the structure of the above-mentioned electronic device. For example, the above-mentioned electronic device can also include more or fewer components than the structure shown in Figure 3 , or have a different configuration from the structure shown in Figure 3 .

[0037] In particular, the processor 110 is configured to provide computing and control capabilities to control the electronic device 100 to perform corresponding tasks, e.g., to control the electronic device 100 to perform any of the vehicle tracking methods provided by the embodiments of the present application, or to perform steps in any of the possible implementation manners of any of the vehicle tracking methods provided by the embodiments of the present application. Those skilled in the art can understand that the processor 110 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; and can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component.

[0038] The memory 120, as a non-transitory computer readable storage medium, can be configured to store non-transitory software programs, non-transitory computer executable programs, instructions and modules, e.g., programs, instructions and modules corresponding to the vehicle tracking methods in the embodiments of the present application. In some embodiments, the memory 120 can include a program storage area and a data storage area, the program storage area can store an operating system and application programs required by at least one function, and the data storage area can store data created by the processor 110, etc. The processor 110 performs various functional applications and data processing of the electronic device 100 by running the non-transitory software programs, instructions and modules stored in the memory 120, so as to implement any of the vehicle tracking methods provided by the embodiments of the present application, or to perform steps in any of the possible implementation manners of any of the vehicle tracking methods provided by the embodiments of the present application. In some embodiments, the memory 120 can include a high-speed random access memory, and can also include a non-transitory memory, e.g., at least one magnetic disk storage device, a flash memory device or other non-transitory solid-state memory device. In some embodiments, the memory 120 can also include a memory remotely arranged with respect to the processor 110, and these remotely arranged memories can be connected to the processor 110 through a communication network. It can be understood that examples of the above communication network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network and combinations thereof.

[0039] According to the above, it can be understood that the implementation subject of any one of the vehicle tracking methods provided by the embodiments of the present application can be any suitable type of electronic device with certain computing and control capabilities, for example, can be implemented by the electronic device 100 described above. In some possible implementations, any one of the vehicle tracking methods provided by the embodiments of the present application can be implemented by a processor executing computer program instructions stored in a memory.

[0040] The vehicle tracking method provided by the embodiments of the present application will be described in detail below in combination with an exemplary application and implementation of the electronic device provided by the embodiments of the present application.

[0041] Please refer to Figure 4 , Figure 4 The flowchart schematically shows the vehicle tracking method provided by some embodiments of the present application.

[0042] Those skilled in the art can understand that the vehicle tracking method provided by the embodiments of the present application can be applied to the electronic device (for example, the electronic device 100) described above. Specifically, the implementation subject of the vehicle tracking method is one or at least two processors of the electronic device.

[0043] As Figure 4 shown, the vehicle tracking method includes but is not limited to the following steps S100-S500: S100: Obtain vehicle image data.

[0044] In the embodiments of the present application, the vehicle image data includes a plurality of original images, and the plurality of original images are images obtained by a plurality of camera devices arranged at different spatial positions at the same time.

[0045] For example, after the plurality of camera devices arranged at different spatial positions obtain the original images at the same time, the original images obtained by each camera are transmitted to the electronic device through a network, so that the electronic device obtains the plurality of original images, that is, obtains the vehicle image data. In some embodiments, after obtaining the vehicle image data, the vehicle image data is preprocessed, such as denoising, image enhancement, normalization, etc., to obtain image data with higher quality.

[0046] It can be understood that there are many ways to obtain vehicle image data. In some embodiments, the electronic device obtains the original images transmitted directly by the camera device. In some embodiments, the original images obtained by the plurality of camera devices are saved in a memory card, and when the memory card is connected to the electronic device, the electronic device reads the original images saved in the memory card. In some embodiments, the original images are saved in advance in a server in the cloud, and the electronic device requests the server in the cloud to return the original images. In some embodiments, the original images are saved in advance in the local storage of the electronic device, and the electronic device directly retrieves the original images from the local storage.

[0047] It is easily understood that other any suitable way can be adopted to obtain the vehicle image data, and the embodiments of the present application do not make any limitation in this regard.

[0048] S200: Based on the plurality of original images, the driving data and the feature data of the vehicle are extracted.

[0049] In the present embodiment, the driving data is data for representing the driving state of the vehicle, and the driving data includes but is not limited to the speed, acceleration, driving direction, initial position and displacement trajectory of the vehicle. The feature data is data for representing the appearance features of the vehicle, and the feature data includes but is not limited to the color, size, shape, model, license plate, roof logo and other appearance information of the vehicle.

[0050] Specifically, a target detection algorithm (such as YOLO, Faster R-CNN) is used to detect the target of each original image, and the position of each target object in each original image is obtained. It can be understood that the target object includes a vehicle and a non-vehicle. The target object region (Bounding Box) is extracted, and the target object region is segmented / masked to obtain the size, shape and other information of the target object. Then the dominant color and color histogram of the target object are extracted to obtain the color information of the target object. The neural network (such as ResNet) and feature extraction algorithm (such as SIFT, ORB) are used to extract the deep feature vector and detailed information, and the structural features and local texture features of the target object are obtained. The characters in the target object are extracted using OCR technology to obtain the identification information of the target object. When the target object is a vehicle, the identification information is the license plate of the vehicle, the slogan attached to the vehicle, etc.

[0051] Among them, the driving data reflects the motion information of the vehicle in the three-dimensional space, such as position, speed, acceleration, driving direction, etc. The three-dimensional estimation is combined with time and the positions of the plurality of camera devices. Specifically, the plurality of original images taken by different camera devices at the same time are three-dimensionally reconstructed, the intrinsic and extrinsic parameters of the camera devices and the positions of the target objects in the original images are combined, and the 3D coordinates (i.e. the initial position) of the target objects are calculated by geometric back-projection. Based on the three-dimensional position changes of the target objects in the plurality of original images, the speed vector and the acceleration vector are calculated to obtain the speed and acceleration of the target objects. The motion vector of the pixel points is calculated by the Hough transform and the optical flow method to determine the motion direction of the target objects. The dynamic time warping (DTW) and hidden Markov model (HMM) algorithms are used to analyze the continuous motion trajectory of the target objects to obtain the displacement trajectory of the target objects.

[0052] S300: Based on the driving data and the feature data, the similarity score of the candidate vehicle in the candidate original image is calculated.

[0053] In the embodiment, the similarity score represents the similarity between the candidate vehicle and the reference vehicle in the reference original image, and the candidate original image and the reference original image are any two original images in the plurality of original images, the candidate vehicle is any one vehicle in the candidate original image, and the reference vehicle is any one vehicle in the reference original image.

[0054] Specifically, any two original images in the plurality of original images are selected as the candidate original image and the reference original image respectively, any one vehicle in the reference original image is selected as the reference vehicle, and any one vehicle in the candidate original image is selected as the candidate vehicle. The similarity score of the candidate vehicle is calculated based on the driving data and the feature data of the reference vehicle and the driving data and the feature data of the candidate vehicle.

[0055] In the embodiment, the driving data and the feature data of all candidate vehicles in the candidate original image are compared and calculated with the driving data and the feature data of the reference vehicle respectively to obtain the similarity scores of all candidate vehicles.

[0056] For example, referring to Figure 2a and Figure 2b , the first original image 201 and the second original image 202 are selected as the reference original image and the candidate original image respectively, any one vehicle (for example, vehicle 12) in the first original image 201 is selected as the reference vehicle, and any one vehicle (for example, vehicle 12, vehicle 14, tree 15, tree 16, vehicle 17 and vehicle 18) in the second original image 202 is selected as the candidate vehicle. Then, the driving data and the feature data of all candidate vehicles in the second original image 202 are compared and calculated with the driving data and the feature data of the reference vehicle respectively to obtain the similarity scores of all candidate vehicles. After the similarity scores of all candidate vehicles in the candidate original image are calculated when the selected vehicle 12 is used as the reference vehicle, other vehicles in the first original image 201 are selected as the reference vehicle, and the driving data and the feature data of all candidate vehicles in the second original image 202 are compared and calculated with the driving data and the feature data of the selected other vehicles as the reference vehicle to obtain the similarity scores of all candidate vehicles in the second original image 202. The above steps are repeatedly executed until the similarity scores of all candidate vehicles in the second original image 202 are calculated when all vehicles in the first original image 201 are used as the reference vehicle. Then, any two original images in the plurality of original images are selected as the reference original image and the candidate original image respectively, and the similarity scores of all candidate vehicles in the candidate original image are calculated.

[0057] In some embodiments, the similarity score of the candidate vehicle in the candidate original image is calculated based on the driving data and the feature data, specifically including but not limited to the following steps S310-S340: S310: Obtaining a confidence score of the candidate vehicle based on the feature data of the candidate vehicle.

[0058] In this embodiment, the driving parameters and appearance parameters of the vehicle can be pre-set. When using a target detection algorithm (such as YOLO or Faster R-CNN) to detect targets in each original image, the target detection algorithm filters out vehicle targets in the original image based on the vehicle's driving parameters and appearance parameters, and eliminates non-vehicle targets in the original image, thereby obtaining feature data of all vehicles in each original image. The target detection algorithm calculates a confidence score for each vehicle based on the vehicle's feature data, and outputs the confidence score for each vehicle in each original image, that is, the credibility of each vehicle as a real vehicle.

[0059] For example, in some embodiments, obtaining a confidence score of a candidate vehicle based on the feature data of the candidate vehicle specifically includes but is not limited to the following steps S311: S311: Compare the size, shape and color of the candidate vehicle with the standard size, standard shape and standard color of the preset vehicle model respectively to obtain a first comparison result.

[0060] In this embodiment of the present invention, the characteristic data includes the size, shape, and color of the vehicle. Preset vehicle models are pre-set by the designer based on experimental data and historical experience. They are used to represent various types of vehicles with standard appearance characteristics (e.g., standard size, standard shape, and standard color).

[0061] Specifically, for each candidate original image, the size of the candidate vehicle in the candidate original image is compared and calculated with the standard size of the preset vehicle model, and the shape of the candidate vehicle in the candidate original image is compared and calculated with the standard shape of the preset vehicle model, and the color of the candidate vehicle in the candidate original image is compared and calculated with the standard color of the preset vehicle model, to obtain multiple first comparison results, and the first comparison result is the confidence score, wherein the confidence score is used to characterize the credibility of the candidate vehicle being a real vehicle, that is, the possibility that the candidate vehicle is a real vehicle.

[0062] For example, assume that the candidate original image is Figure 2b In the second original image 202 shown, the size, shape, and color of each candidate vehicle (i.e., vehicle 12, vehicle 14, vehicle 17, and vehicle 18) in the second original image 202 are compared and calculated with the standard size, standard shape, and standard color of the preset vehicle model, respectively, to obtain multiple first comparison results. Each first comparison result corresponds to a candidate vehicle, and the first comparison result is a confidence score, that is, a confidence score for each candidate vehicle in the second original image 202 is obtained.

[0063] S320: Obtain a matching score between the candidate vehicle and the reference vehicle based on the feature data of the candidate vehicle and the reference vehicle.

[0064] In this embodiment, the feature data of the candidate vehicle is compared and calculated with the feature data of the reference vehicle to obtain a matching score between the candidate vehicle and the reference vehicle. For example, for each candidate original image, the feature data of each candidate vehicle in the candidate original image is compared and calculated with the feature data of the reference vehicle to obtain a matching score between each candidate vehicle and the reference vehicle, that is, the degree of similarity in appearance features between each candidate vehicle and the reference vehicle.

[0065] In some embodiments, based on the feature data of the candidate vehicle and the reference vehicle, obtaining a matching score between the candidate vehicle and the reference vehicle specifically includes but is not limited to the following steps S321: S321: Compare the size, shape, color, vehicle logo and license plate of the candidate vehicle with the reference size, reference shape, reference color, reference vehicle logo and reference license plate of the reference vehicle to obtain a second comparison result.

[0066] In this embodiment, the characteristic data also includes the vehicle logo and license plate. In some embodiments, the vehicle logo is also represented as a vehicle model to characterize the style and model of the vehicle.

[0067] Specifically, for each candidate original image, the size of the candidate vehicle in the candidate original image is compared and calculated with the reference size of the reference vehicle, and the shape of the candidate vehicle in the candidate original image is compared and calculated with the reference shape of the reference vehicle, and the color of the candidate vehicle in the candidate original image is compared and calculated with the reference color of the reference vehicle, and the logo of the candidate vehicle in the candidate original image is compared and calculated with the reference logo of the reference vehicle, and the license plate of the candidate vehicle in the candidate original image is compared and calculated with the reference license plate of the reference vehicle, to obtain multiple second comparison results, and the second comparison result is a matching score, where the matching score is used to characterize the matching degree between the candidate vehicle and the reference vehicle, that is, the degree of similarity between the candidate vehicle and the reference vehicle in appearance features.

[0068] For example, assume that the candidate original image is Figure 2b In the second original image 202 shown, the size, shape, color, vehicle logo, and license plate of each candidate vehicle (i.e., vehicle 12, vehicle 14, vehicle 17, and vehicle 18) in the second original image 202 are compared and calculated with the reference size, reference shape, reference color, reference vehicle logo, and reference license plate of the reference vehicle, respectively, to obtain multiple second comparison results, each of which corresponds to a candidate vehicle. The second comparison results are matching scores, i.e., matching scores between each candidate vehicle in the second original image 202 and the reference vehicle are obtained.

[0069] S330: Obtain the motion trajectory and the stability score of the candidate vehicle based on the driving data of the candidate vehicle and the reference vehicle.

[0070] In the embodiments of the present application, the spatial positions of the candidate vehicle and the reference vehicle in the plurality of original images are connected to form a sequence of trajectory points, Kalman filter or Bezier curve is used for trajectory fitting and smoothing processing to obtain the motion trajectory of the candidate vehicle. Dynamic time warping (DTW) and hidden Markov model (HMM) algorithm are used to analyze and process the continuous motion trajectory of the candidate vehicle, and the stability score of the motion trajectory of the candidate vehicle is calculated according to the speed, direction and motion trajectory of the adjacent trajectory points, whether the motion trajectory is located in the legal lane, the driving area, etc. The stability score is used to represent the stability of the motion trajectory.

[0071] In some embodiments, the motion trajectory and the stability score of the candidate vehicle are obtained based on the driving data of the candidate vehicle and the reference vehicle, specifically including but not limited to the following steps S331: S331: Calculate the motion trajectory and the stability score of the motion trajectory of the candidate vehicle based on the initial position, speed, acceleration and driving direction of the candidate vehicle, and the reference initial position, reference speed, reference acceleration and reference driving direction of the reference vehicle.

[0072] In the embodiments, the driving data includes the initial position, speed, acceleration and driving direction of the vehicle.

[0073] Specifically, the first reference initial position of the candidate vehicle is calculated by using Kalman filtering algorithm according to the initial position, speed, acceleration and driving direction of the candidate vehicle, and the second reference initial position of the reference vehicle is calculated by using Kalman filtering algorithm according to the reference initial position, reference speed, reference acceleration and reference driving direction of the reference vehicle. The first reference initial position and the second reference initial position of the candidate vehicle and the reference vehicle in the plurality of original images are connected to obtain the motion trajectory of the candidate vehicle. Dynamic time warping (DTW) and hidden Markov model (HMM) algorithm are used to analyze and process the motion trajectory of the candidate vehicle, that is, the stability score of the motion trajectory of the candidate vehicle is calculated according to whether the change of the speed, acceleration and driving direction of the adjacent trajectory points is continuous and stable, whether the motion trajectory is located in the legal lane, the driving area and the deviation of the motion trajectory from the driving direction of the vehicle, etc.

[0074] For example, the candidate original image is Figure 2bThe second original image 202 is shown, based on the initial position, speed, acceleration and driving direction of each candidate vehicle (i.e. vehicle 12, vehicle 14, vehicle 17 and vehicle 18) in the second original image 202, and the reference initial position, reference speed, reference acceleration and reference driving direction of the reference vehicle, the motion trajectory of each candidate vehicle and the stability score of the motion trajectory are obtained.

[0075] S340: Based on the confidence score, the matching degree score and the stability score, the similarity score of the candidate vehicle is obtained.

[0076] In some embodiments, the confidence score, the stability score and the matching degree score of the candidate vehicle are added to obtain the similarity score of the candidate vehicle.

[0077] Of course, other any suitable processing can also be performed on the confidence score, the matching degree score and the stability score of the candidate vehicle to obtain the similarity score of the candidate vehicle. For example, the confidence score, the matching degree score and the stability score of the candidate vehicle can also be weighted and fused to obtain the similarity score of the candidate vehicle.

[0078] In some embodiments, based on the confidence score, the matching degree score and the stability score, the similarity score of the candidate vehicle is obtained, specifically including but not limited to the following step S341: S341: The confidence score, the matching degree score and the stability score are weighted and fused to obtain the similarity score of the candidate vehicle.

[0079] In this embodiment, a score weighted fusion formula is designed in advance, and the fusion weight coefficients of the confidence score, the matching degree score and the stability score in the score weighted fusion formula are set according to the actual requirements and the characteristics of the data, so as to obtain the most accurate similarity score of the candidate vehicle. For example, in some embodiments, the score weighted fusion formula is: wherein is the similarity score of the candidate vehicle, , and are the confidence score, the matching degree score and the stability score, respectively, , and are the fusion weight coefficients of the confidence score, the matching degree score and the stability score, respectively.

[0080] Specifically, after the score weighted fusion formula is designed, the confidence score, the matching degree score and the stability score are substituted into the score weighted fusion formula, and the confidence score, the matching degree score and the stability score are weighted and fused to obtain the similarity score of the candidate vehicle.

[0081] S400: determining the target vehicle based on the similarity score.

[0082] In this embodiment, the target vehicle is the candidate vehicle corresponding to the similarity score meeting the preset score requirement, and the preset score requirement can be that the similarity score meets a preset threshold, for example, the similarity score is greater than a similarity score threshold and is the maximum value among the similarity scores of all candidate vehicles (i.e., the similarity score is the maximum similarity score).

[0083] Specifically, the similarity score greater than the similarity score threshold and being the maximum value among the similarity scores of all candidate vehicles is selected from the similarity scores of all candidate vehicles as the target similarity score, and the candidate vehicle corresponding to the target similarity score is determined as the target vehicle, wherein the target vehicle and the reference vehicle are the same vehicle.

[0084] Exemplarily, in some embodiments, the target vehicle is determined based on the similarity score, specifically including but not limited to the following steps S410-S420: S410: selecting a similarity score greater than a similarity score threshold and being the maximum value among the similarity scores as a target similarity score.

[0085] S420: determining the candidate vehicle corresponding to the target similarity score as the target vehicle.

[0086] Exemplarily, each of the plurality of similarity scores is compared with the similarity score threshold to screen one or more reference similarity scores greater than the similarity score threshold, each of the one or more reference similarity scores is compared to screen the maximum reference similarity score (i.e., the maximum value among the one or more reference similarity scores) among the one or more reference similarity scores, and the maximum reference similarity score is selected as the target similarity score, i.e., the similarity score greater than the similarity score threshold and being the maximum value among the plurality of similarity scores is selected as the target similarity score.

[0087] After the target similarity score is determined, the candidate vehicle corresponding to the target similarity score is determined as the target vehicle, and the target vehicle and the reference vehicle are the same vehicle.

[0088] S500: obtaining the initial positions of the target vehicle and the reference vehicle from the driving data, and obtaining the target position based on the initial positions of the target vehicle and the reference vehicle.

[0089] In this embodiment, the target position represents the positions of the target vehicle and the reference vehicle.

[0090] Specifically, an initial position of the target vehicle is obtained from the driving data of the target vehicle, an initial position of the reference vehicle is obtained from the driving data of the reference vehicle, the initial position of the target vehicle and the initial position of the reference vehicle are fused to obtain a target position, and the target position is more accurate position information of the target vehicle and the reference vehicle.

[0091] For example, in some embodiments, the target position is obtained based on the initial position of the target vehicle and the initial position of the reference vehicle, and specifically includes but is not limited to the following steps S510-S560: S510: Construct a weighted fusion formula based on the target initial position and the reference initial position.

[0092] In this step, the original image in which the target vehicle is located is a target original image, and each camera corresponds to an original image.

[0093] In this embodiment, the target initial position and the reference initial position are the initial position of the target vehicle and the initial position of the reference vehicle respectively, and the weighted fusion formula includes the target initial position, the reference initial position, a weight coefficient of the target initial position, and a weight coefficient of the reference initial position.

[0094] In the embodiment of the application, the weighted fusion formula is constructed in advance according to the target initial position and the reference initial position, and the weight coefficient of the target initial position and the weight coefficient of the reference initial position in the weighted fusion formula are set according to actual requirements and characteristics of data. For example, in some embodiments, the weighted fusion formula is:

[0095] wherein, is the target position, is the initial position of the target vehicle in the target original image (i.e., the target initial position) or the initial position of the reference vehicle in the reference original image (i.e., the reference initial position), is the weight coefficient of the target initial position or the weight coefficient of the reference initial position.

[0096] S520: Obtain the target extrinsic calibration accuracy of the target camera, the target resolution, and the target picture position of the target vehicle in the target original image.

[0097] In this embodiment, the target camera is the camera corresponding to the target original image, i.e., the camera that captures the target original image.

[0098] Specifically, the target extrinsic calibration accuracy and the target resolution of the target camera are obtained from the target camera, and the target picture position of the target vehicle in the target original image is obtained by detecting the target original image using a target detection algorithm. The target resolution refers to the resolution of the original image captured by the target camera.

[0099] S530: Obtain the reference extrinsic calibration accuracy, the reference resolution, and the reference picture position of the reference vehicle in the reference original image of the reference camera.

[0100] In this embodiment, the reference camera is the camera corresponding to the reference original image, i.e., the camera that captures the reference original image.

[0101] Specifically, the reference extrinsic calibration accuracy, the reference resolution of the reference camera are obtained, and the reference picture position of the reference vehicle in the reference original image is obtained by detecting the reference original image using a target detection algorithm. The reference resolution refers to the resolution of the original image captured by the reference camera.

[0102] S540: Calculate the weight coefficient of the target initial position based on the target extrinsic calibration accuracy, the target resolution, and the target picture position.

[0103] In this step, a weight coefficient calculation formula is designed in advance, i.e., according to the target extrinsic calibration accuracy, the target resolution, and the target picture position, the weight coefficients corresponding to the target extrinsic calibration accuracy, the target resolution, and the target picture position are set, and a first weight coefficient calculation formula is constructed according to the weight coefficients corresponding to the target extrinsic calibration accuracy, the target resolution, and the target picture position.

[0104] For example, in some embodiments, the first weight coefficient calculation formula is:

[0105] wherein, is the weight coefficient of the target initial position, is the weight coefficient corresponding to the target extrinsic calibration accuracy, is the weight coefficient corresponding to the target resolution, is the weight coefficient corresponding to the target picture position.

[0106] In this embodiment, when the target extrinsic calibration accuracy is meter-level (m-level) accuracy, the weight coefficient corresponding to the target extrinsic calibration accuracy (i.e., w1) is determined to be 0.3, and correspondingly, when the target extrinsic calibration accuracy is centimeter-level (cm-level) accuracy, the weight coefficient corresponding to the target extrinsic calibration accuracy (i.e., w1) is determined to be 0.7. In this embodiment, when the target resolution is 1080P or above, the weight coefficient corresponding to the target resolution (i.e., w2) is determined to be 0.6, and when the target resolution is below 1080P, the weight coefficient corresponding to the target resolution (i.e., w2) is determined to be 0.4.

[0107] In this embodiment, when the target resolution is 1080P or above, the weight coefficient corresponding to the target resolution (i.e., w2) is determined to be 0.6, and when the target resolution is below 1080P, the weight coefficient corresponding to the target resolution (i.e., w2) is determined to be 0.4. ​​​

[0108] In this embodiment, the target picture position of the target vehicle in the target original image refers to the area position of the target vehicle in the target original image. The target original image is divided into a plurality of sub-picture images in the longitudinal direction. When the target vehicle is in one or more target sub-picture images close to the target camera, the weight coefficient corresponding to the target picture position (i.e. ) is determined to be 0.7. When the target vehicle is in one or more target sub-picture images away from the target camera, the weight coefficient corresponding to the target picture position (i.e. ) is determined to be 0.3.

[0109] For example, referring to FIG. 2, Figure 5 , Figure 5 the target sub-picture images in the second original image 202 include target sub-picture image 2021, target sub-picture image 2022, target sub-picture image 2023, target sub-picture image 2024, target sub-picture image 2025, and target sub-picture image 2026. When the target vehicle is in target sub-picture image 2024, target sub-picture image 2025, and / or target sub-picture image 2026 close to the target camera, the weight coefficient corresponding to the target picture position is determined to be 0.7. When the target vehicle is in target sub-picture image 2021, target sub-picture image 2022, and / or target sub-picture image 2023 away from the target camera, the weight coefficient corresponding to the target picture position is determined to be 0.3.

[0110] Specifically, after determining the weight coefficients corresponding to the target extrinsic calibration accuracy, the target resolution, and the target picture position, the weight coefficients corresponding to the target extrinsic calibration accuracy, the target resolution, and the target picture position are substituted into the weight coefficient calculation formula to calculate the weight coefficient of the target initial position.

[0111] S550: Calculate the weight coefficient of the reference initial position based on the reference extrinsic calibration accuracy, the reference resolution, and the reference picture position.

[0112] In this step, the weight coefficient calculation formula is designed in advance, that is, the weight coefficients corresponding to the reference extrinsic calibration accuracy, the reference resolution, and the reference picture position are set according to the reference extrinsic calibration accuracy, the reference resolution, and the reference picture position, and the second weight coefficient calculation formula is constructed according to the weight coefficients corresponding to the reference extrinsic calibration accuracy, the reference resolution, and the reference picture position.

[0113] For example, in some embodiments, the second weight coefficient calculation formula is:

[0114] wherein, is the weight coefficient of the reference initial position, is the weight coefficient corresponding to the reference extrinsic calibration accuracy, a weight coefficient corresponding to a reference resolution, a weight coefficient corresponding to a reference picture position.

[0115] In this embodiment, when the reference extrinsic calibration accuracy is meter-level (m-level) accuracy, the weight coefficient (i.e., w1) corresponding to the reference extrinsic calibration accuracy is determined to be 0.3. Correspondingly, when the reference extrinsic calibration accuracy is centimeter-level (cm-level) accuracy, the weight coefficient (i.e., w1) corresponding to the reference extrinsic calibration accuracy is determined to be 0.7.

[0116] In this embodiment, when the reference resolution is 1080P or above, the weight coefficient (i.e., w2) corresponding to the reference resolution is determined to be 0.6. When the reference resolution is below 1080P, the weight coefficient (i.e., w2) corresponding to the reference resolution is determined to be 0.4.

[0117] In this embodiment, the reference picture position of the reference vehicle in the reference original image refers to the area position of the reference vehicle in the reference original image. When the reference vehicle is in one or more reference sub-picture images close to the reference camera device, the weight coefficient (i.e., w3) corresponding to the reference picture position is determined to be 0.7. When the reference vehicle is in one or more reference sub-picture images away from the reference camera device, the weight coefficient (i.e., w3) corresponding to the reference picture position is determined to be 0.3.

[0118] For example, as shown in FIG. 2, the reference sub-picture images in the second original image 201 include reference sub-picture image 2011, reference sub-picture image 2012, reference sub-picture image 2013, reference sub-picture image 2014, reference sub-picture image 2015, and reference sub-picture image 2016. When the reference vehicle is in the reference sub-picture image 2014, the reference sub-picture image 2015, and / or the reference sub-picture image 2016 close to the reference camera device, the weight coefficient corresponding to the reference picture position is determined to be 0.7. When the reference vehicle is in the reference sub-picture image 2011, the reference sub-picture image 2012, and / or the reference sub-picture image 2013 away from the reference camera device, the weight coefficient corresponding to the reference picture position is determined to be 0.3. Figure 6 Figure 6

[0119] Specifically, after determining the weight coefficients corresponding to the reference extrinsic calibration accuracy, the reference resolution, and the reference picture position, the weight coefficients corresponding to the reference extrinsic calibration accuracy, the reference resolution, and the reference picture position are substituted into the weight coefficient calculation formula to calculate the weight coefficient of the reference initial position.​​​​​​​​

[0120] S560: Substituting the target initial position and the reference initial position into the weighted fusion formula to obtain the target position.

[0121] Specifically, after calculating the weight coefficient of the target initial position and the weight coefficient of the reference initial position in the weighted fusion formula, the target initial position and the reference initial position are substituted into the weighted fusion formula, and the target position is obtained through calculation of the weighted fusion formula. The target position is more accurate position information of the target vehicle and the reference vehicle.

[0122] In summary, the vehicle tracking method provided by the embodiment of the application comprises: obtaining vehicle image data, the vehicle image data comprising a plurality of original images, the plurality of original images being images obtained by a plurality of camera devices arranged at different spatial positions at the same time, extracting driving data and feature data of a vehicle in the original images, the driving data being data for representing a driving state of the vehicle, and the feature data being data for representing an appearance feature of the vehicle, calculating a similarity score of a candidate vehicle in a candidate original image based on the driving data and the feature data, the similarity score representing a similarity between the candidate vehicle and a reference vehicle in a reference original image, the candidate original image and the reference original image being any two original images in the plurality of original images, the candidate vehicle being any one vehicle in the candidate original image, and the reference vehicle being any one vehicle in the reference original image, determining a target vehicle based on the similarity score, the target vehicle being a candidate vehicle corresponding to a similarity score meeting a preset score requirement, the target vehicle and the reference vehicle being the same vehicle, obtaining initial positions of the target vehicle and the reference vehicle from the driving data, and obtaining a target position based on the initial positions of the target vehicle and the reference vehicle, the target position representing a position of the target vehicle and the reference vehicle.

[0123] The embodiment of the application extracts driving data and feature data of a vehicle from a plurality of original images captured by a plurality of camera devices at the same time, determines the same vehicle in the plurality of original images based on the driving data and the feature data, and obtains more accurate position information of the vehicle according to initial position information of the same vehicle in the plurality of original images. In this way, the vehicle can be continuously and accurately tracked, and the accuracy of vehicle tracking is improved.

[0124] The embodiment of the application provides a computer readable storage medium, the computer readable storage medium storing computer program instructions executable by a processor, the computer program instructions being executed by the processor to cause the computer to execute any one of the vehicle tracking methods provided by the embodiment of the application, or execute steps in any one of the possible implementation manners of the vehicle tracking methods provided by the embodiment of the application.

[0125] In some embodiments, the storage medium can be flash memory, hard disk, optical disk, register, magnetic surface memory, removable disk, CD-ROM, random access memory (RAM), read only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, and other memory known in the art, or any other form of storage medium known in the art, or various devices including one or any combination of the above storage media.

[0126] In some embodiments, the computer program instructions can be in the form of programs, software, software modules, scripts or code, written in any form of programming language, including compiled or interpreted languages, or declarative or procedural languages, and can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0127] By way of example, computer program instructions can be, but are not limited to, correspond to a file in a file system, can be stored in a portion of a file that holds other programs or data, for example, one or more scripts stored in a HTML (Hyper Text Markup Language) document, in a single file dedicated to the program in question, or in multiple coordinated files, for example, files that store one or more modules, sub programs, or portions of code.

[0128] By way of example, computer program instructions can be deployed to execute on one computer (including devices such as smart terminals and servers), or on multiple computers located at one site, or on multiple computers distributed across multiple sites and interconnected by a communication network. It is readily understood that all or part of the methods described in the embodiments of the present application can be implemented directly using electronic hardware or computer program instructions executable by a processor, or a combination of the two.

[0129] Those skilled in the art can understand that the embodiments provided by the present application are only illustrative, and the writing order of each step in the method of the embodiments does not mean a strict execution order and constitutes any limitation on the implementation process, and the order can be adjusted, combined and deleted according to actual needs. The modules or sub modules, units or sub units, etc. in the device or system of the embodiments can be combined, divided and deleted according to actual needs. For example, the division of the unit is only a logical functional division, and another division mode can also be used in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed.

[0130] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, and of course can also be implemented by hardware. Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments.

[0131] It should be noted that the above embodiments are intended to illustrate the technical concepts and characteristics of the present application, and the purpose is to enable those skilled in the art to understand the content of the present application and implement it accordingly, and cannot be used to limit the scope of protection of the present application. Those skilled in the art can understand that all or part of the processes of the above-mentioned embodiments can be modified according to the technical solutions described in the embodiments of the present application, or some technical features can be replaced equivalently. It can be understood that these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should be regarded as equivalent changes and modifications based on the embodiments of the present application, and should belong to the scope of the claims of the present application.

Claims

1. A vehicle tracking method, characterized in that: include: Acquire vehicle image data, where the vehicle image data includes a plurality of original images, where the plurality of original images are images captured at the same time by a plurality of cameras arranged at different spatial positions; Extracting driving data and feature data of the vehicle based on the plurality of original images, wherein the driving data is data used to characterize the driving state of the vehicle, and the feature data is data used to characterize the appearance characteristics of the vehicle; calculating, based on the driving data and the feature data, a similarity score for a candidate vehicle in a candidate original image, the similarity score representing a similarity between the candidate vehicle and a reference vehicle in a reference original image, the candidate original image and the reference original image being any two original images from a plurality of original images, the candidate vehicle being any one of the candidate original images, and the reference vehicle being any one of the reference original images; Determining a target vehicle based on the similarity score, the target vehicle being a candidate vehicle corresponding to a similarity score that meets a preset score requirement, and the target vehicle and the reference vehicle being the same vehicle; Initial positions of the target vehicle and the reference vehicle are acquired from the driving data, and a target position is obtained based on the initial positions of the target vehicle and the reference vehicle, wherein the target position represents the positions of the target vehicle and the reference vehicle.

2. The method according to claim 1, characterized in that The calculating, based on the driving data and the feature data, a similarity score of the candidate vehicle in the candidate original image includes: Obtaining a confidence score for the candidate vehicle based on the feature data of the candidate vehicle; Obtaining a matching score between the candidate vehicle and the reference vehicle based on the feature data of the candidate vehicle and the reference vehicle; Obtaining a motion trajectory and a stability score of the candidate vehicle based on the driving data of the candidate vehicle and the reference vehicle, wherein the stability score is used to characterize the stability of the motion trajectory; A similarity score of the candidate vehicle is obtained based on the confidence score, the matching score, and the stability score.

3. The method according to claim 2, characterized in that The characteristic data includes size, shape, and color. Obtaining a confidence score of the candidate vehicle based on the characteristic data of the candidate vehicle includes: The size, shape and color of the candidate vehicle are respectively compared with the standard size, standard shape and standard color of the preset vehicle model to obtain a first comparison result, which is the confidence score, and the confidence score is used to characterize the credibility of the candidate vehicle being a real vehicle.

4. The method according to claim 3, characterized in that The feature data also includes a vehicle logo and a license plate. Obtaining a matching score between the candidate vehicle and the reference vehicle based on the feature data of the candidate vehicle and the reference vehicle includes: The size, shape, color, logo and license plate of the candidate vehicle are respectively compared with the reference size, reference shape, reference color, reference logo and reference license plate of the reference vehicle to obtain a second comparison result, which is the matching score, wherein the matching score is used to characterize the matching degree between the candidate vehicle and the reference vehicle.

5. The method according to claim 2, characterized in that The driving data includes an initial position, speed, acceleration, and driving direction. The obtaining of the motion trajectory and stability score of the candidate vehicle based on the driving data of the candidate vehicle and the reference vehicle includes: Based on the initial position, speed, acceleration and driving direction of the candidate vehicle and the reference initial position, reference speed, reference acceleration and reference driving direction of the reference vehicle, a motion trajectory of the candidate vehicle and a stability score of the motion trajectory are calculated.

6. The method according to claim 2, characterized in that The obtaining of the similarity score of the candidate vehicle based on the confidence score, the matching score, and the stability score includes: The confidence score, the matching score, and the stability score are weighted and fused to obtain a similarity score for the candidate vehicle.

7. The method according to any one of claims 1 to 6, characterized in that Determining a target vehicle based on the similarity score includes: Selecting a similarity score among the plurality of similarity scores that is greater than a preset score threshold and is the maximum value among the plurality of similarity scores as a target similarity score; The candidate vehicle corresponding to the target similarity score is determined as the target vehicle.

8. The method according to any one of claims 1 to 6, characterized in that The original image where the target vehicle is located is the target original image, each camera device corresponds to one original image, and obtaining the target position based on the initial positions of the target vehicle and the reference vehicle includes: Constructing a weighted fusion formula based on a target initial position and a reference initial position, wherein the target initial position and the reference initial position are the initial positions of the target vehicle and the reference vehicle, respectively, and the weighted fusion formula includes the target initial position, the reference initial position, a weight coefficient of the target initial position, and a weight coefficient of the reference initial position; Obtaining target extrinsic parameter calibration accuracy and target resolution of a target camera device and a target image position of the target vehicle in the target original image, wherein the target camera device is a camera device corresponding to the target original image; Obtaining a reference extrinsic calibration accuracy and a reference resolution of a reference camera device and a reference image position of the reference vehicle in the reference original image, wherein the reference camera device is a camera device corresponding to the reference original image; Calculating a weight coefficient of the target initial position based on the target extrinsic parameter calibration accuracy, the target resolution, and the target screen position; Calculating a weight coefficient of the reference initial position based on the reference extrinsic parameter calibration accuracy, the reference resolution, and the reference picture position; Substitute the target initial position and the reference initial position into the weighted fusion formula to obtain the target position.

9. An electronic device, characterized in that: include: a processor and a memory communicatively connected to the processor; The memory stores computer program instructions executable by the processor, and when the computer program instructions are executed by the processor, the electronic device executes the vehicle tracking method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer program instructions executable by a processor. When the computer program instructions are executed by the processor, the computer is enabled to execute the vehicle tracking method according to any one of claims 1 to 8.

Images