Vehicle tracking method, system, device and storage medium based on semi-closed scene

Abstract

This invention provides a vehicle tracking method, system, device, and storage medium based on a semi-enclosed scene. The method includes the following steps: acquiring overhead video using several sensors positioned at a high altitude in the semi-enclosed scene; matching the video frames of the overhead video with a high-precision map of the scene to obtain a local map corresponding to the overhead video; performing image recognition on the video frames of the overhead video to obtain at least one vehicle image region and corresponding vehicle information in the overhead video; obtaining the vehicle's position within the local map based on the position of the vehicle image region in the video frame image; and predicting the trajectory of the target vehicle to reach the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, thereby establishing a mapping relationship between vehicle information and the vehicle's motion trajectory in the high-precision map of the semi-enclosed scene. This invention can accurately and continuously track the trajectories of multiple vehicles in a semi-enclosed scene using a small number of cameras, while significantly reducing hardware costs and computing power requirements.

Description

Technical Field

[0001] This invention relates to the field of vehicle area monitoring, and more specifically, to a vehicle tracking system, method, device, and storage medium based on a semi-enclosed environment. Background Technology

[0002] In semi-enclosed industrial parks or ports, safety considerations often necessitate the identification and monitoring of incoming vehicles. Traditionally, low-position cameras are installed at park entrances or key locations to capture license plates and identify passing vehicles. However, for larger parks or ports, this method is insufficient and cannot effectively cover the entire monitored area. Furthermore, due to the wide dimensions of entrances and exits, or the potential for obstruction by other vehicles, missed or false images can occur, leading to questionable effectiveness. Semi-enclosed environments are characterized by large footprints and diverse driving paths, but a relatively concentrated variety and small number of vehicle types. Even migrating existing monitoring systems to semi-enclosed environments often results in either insufficient recognition accuracy or excessively high computational costs, making accurate implementation impossible.

[0003] In view of this, the present invention provides a vehicle tracking method, system, device and storage medium based on a semi-enclosed scenario.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] To address the problems in the prior art, the present invention aims to provide a vehicle tracking method, system, device, and storage medium based on a semi-enclosed scenario. It overcomes the difficulties of the prior art, enables accurate and continuous tracking of multiple vehicle trajectories in a semi-enclosed scenario using a small number of cameras, and greatly reduces the required hardware costs and computing power.

[0006] Embodiments of the present invention provide a vehicle tracking method based on a semi-enclosed scenario, comprising the following steps:

[0007] A top-down video feed was captured using several sensors positioned at a high vantage point in a semi-enclosed environment.

[0008] The high-precision map of the semi-enclosed scene is matched with the video frames of the overhead video to obtain a local map corresponding to the overhead video.

[0009] Image recognition is performed on the video frames of the overhead view video to obtain at least one vehicle image region and corresponding vehicle information in the overhead view video;

[0010] The position of the vehicle within the local map is obtained based on the position of the vehicle image region in the video frame image; and

[0011] Based on the trajectory of the target vehicle in the local map corresponding to one sensor, the trajectory to the next sensor is predicted, so as to establish a mapping relationship between the vehicle information and the vehicle's motion trajectory in the high-precision map of the semi-enclosed scene.

[0012] Preferably, the step of acquiring top-down video through several sensors positioned at a high camera angle in a semi-enclosed scene includes:

[0013] The sensor has a fixed viewing angle and is installed at several high camera positions within the semi-enclosed scene. The height of these high camera positions above the ground exceeds the maximum height of vehicles allowed to travel within the semi-enclosed scene.

[0014] The sensor acquires overhead video from a top-down perspective.

[0015] Preferably, the step of acquiring top-down video through several sensors positioned at a high camera angle in a semi-enclosed scene includes:

[0016] The sensor has a rotatable viewing angle. After each rotation, image similarity calculation is performed between the video frames of the top-down video and various parts of the high-precision map.

[0017] The local map with the highest image similarity is used as the monitoring range corresponding to the overhead video; and

[0018] Establish a mapping relationship between the pixels representing the ground in the video frame and the positioning information in the high-precision map.

[0019] Preferably, the process of matching the high-precision map based on the semi-enclosed scene with video frames of the overhead view video to obtain a local map corresponding to the overhead view video includes:

[0020] Image similarity calculation is performed between video frames from the overhead view video and various parts of the high-precision map;

[0021] The local map with the highest image similarity is used as the monitoring range corresponding to the overhead video; and

[0022] Establish a mapping relationship between the pixels representing the ground in the video frame and the positioning information in the high-precision map.

[0023] Preferably, the step of performing image recognition on the video frames of the overhead view video to obtain at least one vehicle image region and corresponding vehicle information in the overhead view video includes:

[0024] Image recognition is performed on video frames of the overhead view video using a trained first neural network to obtain at least one vehicle image region in the overhead view video; and

[0025] The vehicle image region is identified by a trained second neural network to obtain vehicle information, which includes at least license plate information.

[0026] Preferably, the vehicle image region is identified by a trained second neural network to obtain vehicle information. The vehicle information includes at least license plate information and also attribute information, which includes at least one of the following: vehicle body color, vehicle type, front shape, rear shape, vehicle markings, and type of vehicle-mounted work arm.

[0027] Preferably, obtaining the position of the vehicle within the local map based on the position of the vehicle image region in the video frame image includes:

[0028] Obtain the coordinates of the center position of the vehicle image region in the image of the video frame; and

[0029] Based on the coordinates of the central location, the corresponding positioning information in the high-precision map is obtained.

[0030] Preferably, the step of predicting the trajectory of the target vehicle to the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, in order to establish a mapping relationship between the vehicle information and the vehicle's motion trajectory in the high-precision map of the semi-enclosed scene, includes:

[0031] When the local maps of the sensors do not overlap, based on the positional relationship between the sensors and their corresponding local maps, the target vehicle's trajectory and speed within a local map corresponding to a sensor, the prediction time for reaching the local map of the next adjacent sensor and the prediction region of the image that preferentially enters the video frame of that adjacent sensor are predicted along the direction of the trajectory; and

[0032] Based on the predicted time, the image in the video frame is identified. When only one vehicle is identified in the predicted area, and the time difference between the arrival time of the vehicle and the predicted time is less than a preset threshold, the motion trajectories under the two sensors are merged as the motion trajectory of the target vehicle. The motion trajectory of the missing area between the motion trajectories is supplemented based on the motion trajectory of the target vehicle under different sensors.

[0033] Preferably, the step of predicting the trajectory of the target vehicle to the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, in order to establish a mapping relationship between the vehicle information and the vehicle's motion trajectory in the high-precision map of the semi-enclosed scene, includes:

[0034] When the local maps of the sensors do not overlap, based on the positional relationship between the sensors and their corresponding local maps, the driving trajectory and speed of the target vehicle in the local map corresponding to one sensor, the prediction time for reaching the local map of the next adjacent sensor is predicted along the direction of the driving trajectory, and the prediction region of the local image of the video frame that preferentially enters the adjacent sensor; and

[0035] Based on the predicted time, identification is performed in the prediction region of the image in the video frame. When only a few vehicles are identified in the prediction region, the graphic difference between the trajectory of the vehicle and the driving trajectory is compared. The vehicle with the smallest graphic difference is taken as the target vehicle. The motion trajectory of the target vehicle is merged, and the motion trajectory of the missing area between the motion trajectories is supplemented based on the motion trajectory of the target vehicle under different sensors.

[0036] Preferably, the step of predicting the trajectory of the target vehicle to the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, in order to establish a mapping relationship between the vehicle information and the vehicle's motion trajectory in the high-precision map of the semi-enclosed scene, includes:

[0037] When the local maps of the sensors do not overlap, based on the positional relationship between the sensors and their corresponding local maps, the target vehicle's trajectory and speed within a local map corresponding to a sensor, the prediction time for reaching the local map of the next adjacent sensor and the prediction region of the image that preferentially enters the video frame of that adjacent sensor are predicted along the direction of the trajectory; and

[0038] Based on the predicted time, identification is performed in the prediction region of the image in the video frame. When only a few vehicles are identified in the prediction region, the similarity of the license plate information and / or attribute information of each vehicle is weighted and summed to obtain a similarity reference value. The vehicle with the largest similarity reference value is taken as the target vehicle. The motion trajectories of the target vehicles are merged, and the motion trajectories of the missing areas between the motion trajectories are supplemented based on the motion trajectories of the target vehicles under different sensors.

[0039] Embodiments of the present invention also provide a vehicle tracking system based on a semi-enclosed scenario, used to implement the above-described vehicle tracking method based on a semi-enclosed scenario. The vehicle tracking system based on a semi-enclosed scenario includes:

[0040] The video acquisition module uses several sensors positioned at a high vantage point in a semi-enclosed environment to capture overhead video.

[0041] The monitoring zoning module matches the high-precision map of the semi-enclosed scene with the video frames of the overhead video to obtain a local map corresponding to the overhead video.

[0042] The image recognition module performs image recognition on the video frames of the overhead view video to obtain at least one vehicle image region and corresponding vehicle information in the overhead view video;

[0043] The map positioning module obtains the position of the vehicle within the local map based on the position of the vehicle image area in the video frame image;

[0044] The motion trajectory module predicts the trajectory of the target vehicle to the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, so as to establish a mapping relationship between the vehicle information and the motion trajectory of the vehicle in the high-precision map of the semi-enclosed scene.

[0045] Embodiments of the present invention also provide a vehicle tracking device based on a semi-enclosed scenario, comprising:

[0046] processor;

[0047] Memory, which stores the processor's executable instructions;

[0048] The processor is configured to execute the steps of the vehicle tracking method based on a semi-closed scenario by executing executable instructions.

[0049] Embodiments of the present invention also provide a computer-readable storage medium for storing a program that, when executed, implements the steps of the vehicle tracking method based on a semi-closed scenario described above.

[0050] The vehicle tracking method, system, device, and storage medium based on semi-enclosed scenarios of the present invention can accurately and continuously track the trajectories of multiple vehicles in a semi-enclosed scenario using a small number of cameras, and greatly reduces the required hardware costs and computing power. Attached Figure Description

[0051] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings.

[0052] Figure 1 This is a flowchart of the vehicle tracking method based on a semi-enclosed scenario according to the present invention.

[0053] Figures 2 to 6 This is a schematic diagram illustrating the process of implementing the vehicle tracking method based on a semi-closed scenario according to the present invention.

[0054] Figure 7 This is a schematic diagram of the vehicle tracking system based on a semi-enclosed scenario according to the present invention.

[0055] Figure 8 This is a structural schematic diagram of the vehicle tracking device based on a semi-enclosed scenario according to the present invention.

[0056] Figure 9 This is a schematic diagram of the structure of a computer-readable storage medium according to an embodiment of the present invention. Detailed Implementation

[0057] The following specific examples illustrate the implementation methods of this application. Those skilled in the art can easily understand the other advantages and effects of this application from the content disclosed herein. This application can also be implemented or applied through other different specific embodiments, and various details in this application can be modified or changed according to different viewpoints and application systems without departing from the spirit of this application. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0058] The embodiments of this application will now be described in detail with reference to the accompanying drawings, so that those skilled in the art can easily implement the application. This application may be embodied in many different forms and is not limited to the embodiments described herein.

[0059] In this application, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics represented in connection with that embodiment or example, which are included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics represented may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate different embodiments or examples represented in this application, as well as features of different embodiments or examples.

[0060] Furthermore, the terms "first" and "second" are used for illustrative purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the representation of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0061] For the purpose of clearly describing this application, devices that are not relevant to the description are omitted, and the same or similar components throughout the specification are given the same reference numerals.

[0062] Throughout this specification, when it is said that a device is "connected" to another device, this includes not only "direct connection" but also "indirect connection" by placing other components in between. Furthermore, when it is said that a device "comprises" a certain constituent element, unless otherwise stated otherwise, this does not exclude other constituent elements, but rather implies that other constituent elements may be included.

[0063] When we say that a device is "above" another device, this can mean that it is directly above the other device, or it can mean that other devices are present in between. Conversely, when we say that a device is "directly" "above" another device, there are no other devices present in between.

[0064] While the terms first, second, etc., are used in some instances to denote various elements in this invention, these elements should not be limited by these terms. These terms are used only to distinguish one element from another. For example, first interface and second interface, etc., are used. Furthermore, as used herein, the singular forms “a,” “an,” and “the” are intended to also include the plural forms, unless the context indicates otherwise. It should be further understood that the terms “comprising,” “including,” indicate the presence of features, steps, operations, elements, components, items, kinds, and / or groups, but do not exclude the presence, occurrence, or addition of one or more other features, steps, operations, elements, components, items, kinds, and / or groups. The terms “or” and “and / or” as used herein are to be interpreted inclusively, or mean any one or any combination thereof. Therefore, “A, B, or C” or “A, B, and / or C” means “any one of: A; B; C; A and B; A and C; B and C; A, B, and C.” An exception to this definition will only occur if the combination of elements, functions, steps, or operations is inherently mutually exclusive in some way.

[0065] The technical terms used herein are for reference only to specific embodiments and are not intended to limit the scope of this application. The singular form used herein includes the plural form unless the statement explicitly indicates otherwise. The word "comprising" as used in the specification means to specify a particular characteristic, region, integer, step, operation, element, and / or component, and does not exclude the presence or addition of other characteristics, regions, integers, steps, operations, elements, and / or components.

[0066] Although not explicitly defined, all terms, including technical and scientific terms used herein, shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. Terms defined in commonly used dictionaries shall be further interpreted as having a meaning consistent with the relevant technical literature and the content of this present application, and shall not be over-interpreted as having an ideal or overly formulaic meaning unless otherwise defined.

[0067] Figure 1 This is a flowchart of the vehicle tracking method based on a semi-enclosed scenario according to the present invention. Figure 1 As shown, an embodiment of the present invention provides a vehicle tracking method based on a semi-enclosed scenario, comprising the following steps:

[0068] S110: Acquires overhead video using several sensors positioned at a high vantage point in a semi-enclosed environment.

[0069] S120. Match the high-precision map of the semi-enclosed scene with the video frames of the overhead video to obtain the local map corresponding to the overhead video.

[0070] S130. Perform image recognition on the video frames of the overhead view video to obtain at least one vehicle image region and corresponding vehicle information in the overhead view video.

[0071] S140. Obtain the position of the vehicle in the local map based on the position of the vehicle image area in the video frame image.

[0072] S150. Based on the trajectory of the target vehicle in the local map corresponding to one sensor, predict the trajectory to the next sensor, so as to establish the mapping relationship between vehicle information and the vehicle's motion trajectory in a high-precision map of a semi-closed scene.

[0073] The semi-enclosed scenarios in this invention refer to scenarios such as unmanned terminals, container yards, and unmanned factories. Unlike common road network scenarios or small, enclosed scenarios, semi-enclosed scenarios are characterized by a large area and diverse driving paths, but the types of vehicles inside are relatively concentrated and the number of vehicle types is not large. Even if existing monitoring systems are migrated to semi-enclosed scenarios, they either cannot achieve the required recognition accuracy or require excessive computing power, resulting in extremely high costs and making accurate implementation impossible. This invention uses cameras and image recognition algorithms to effectively identify and locate multiple types of targets within the site, continuously track different targets, and assign each target a unique number for easy tracking and management. This invention can reuse existing cameras within the site, reducing costs, and can also directly use global coordinate information from high-precision positioning technology, reducing the need for calculating and using topological parameters in camera junction areas.

[0074] The difference between this invention and existing solutions using surveillance cameras positioned at ramps or intersections lies in the following: Surveillance cameras can only collect information on individual vehicles sequentially from their own location at a standard height (1 meter or 1.5 meters) to determine the information of vehicles passing through a ramp or intersection. This results in a very small monitoring range and an inability to track vehicle trajectories within a preset area. In contrast, this invention uses a completely different data source, acquiring video streams from monitors positioned at heights of 10 or 20 meters. This allows for coverage of a larger ground area. Furthermore, by locating and identifying vehicle information in each frame of the image obtained from the high-positioned monitor, a mapping relationship is established between the two, thereby generating specific vehicle movement trajectory information based on the video content, enabling vehicle tracking. Since only a few high-positioned monitors are needed, eliminating the need for numerous surveillance cameras, hardware costs are significantly reduced.

[0075] In a preferred embodiment, step S110 includes:

[0076] S111. Several high-positioned video sensors (e.g., surveillance cameras capturing and transmitting video streams) are installed in a semi-enclosed environment. The height of these high-positioned cameras from the ground is greater than the maximum height of the vehicle in the semi-enclosed environment.

[0077] S112. Acquire overhead video from a top-down perspective using video sensors. In this invention, a high camera position refers to a height greater than 10 meters above the ground. It can be set at a height of 20 meters, 30 meters, 40 meters, 50 meters, etc. The higher the position, the larger the monitoring range, but the more difficult it is to accurately identify vehicle information. However, this is not a limitation.

[0078] In a preferred embodiment, step S120 includes:

[0079] S121. The sensor has a fixed viewing angle and performs image similarity calculations based on video frames from the top-down video and various local areas of the high-precision map.

[0080] S122. Use the local map with the highest image similarity as the monitoring range corresponding to the overhead video.

[0081] as well as

[0082] S123. Establish the mapping relationship between the pixels representing the ground in the video frame and the positioning information in the high-precision map. If the video sensor can be a fixed-viewpoint, then only one preset image comparison algorithm is needed to obtain the positioning for subsequent tracking. In this invention, an existing image comparison algorithm is used to calculate the image similarity between the video frame of the overhead view video and various parts of the high-precision map, thereby finding the local map with the highest similarity to the video frame as the viewpoint area of ​​the overhead view video, but it is not limited to this.

[0083] In a preferred embodiment, step S120 includes:

[0084] S125. The sensor has a rotatable viewing angle. After each rotation, image similarity calculation is performed between the video frames of the top-down video and various parts of the high-precision map.

[0085] S126. Use the local map with the highest image similarity as the monitoring range corresponding to the overhead video.

[0086] as well as

[0087] S127. Establish the mapping relationship between the pixels representing the ground in the video frame and the positioning information in the high-precision map. If the video sensor can have a variable viewing angle (the video sensor can rotate horizontally), then the image comparison algorithm needs to be re-performed for each rotation to obtain the positioning under the current viewing angle for subsequent tracking. This scheme can expand the monitoring range of a single video sensor, but it will sacrifice positioning accuracy and increase some computing power requirements, but this is not the limitation.

[0088] In a preferred embodiment, step S130 includes:

[0089] S131. Image recognition is performed on the video frames of the overhead video using a trained first neural network to obtain at least one vehicle image region in the overhead video.

[0090] S132. Vehicle image regions are identified using a trained second neural network to obtain vehicle information, which includes at least license plate information. Vehicle information also includes attribute information, such as vehicle body color, vehicle type, front shape, rear shape, vehicle markings, and at least one type of onboard work arm. For example, the vehicle type could be a sedan, van, pickup truck, container truck, forklift, stacker, reach stacker, etc. These attributes are bound to each identified vehicle, and a different identification ID is assigned to each different vehicle, but this is not a limitation. In this invention, an existing trained vehicle recognition neural network can be used to perform image recognition on video frames of a top-down video, thereby obtaining at least one vehicle image region in the video frame. Other neural networks used for vehicle type classification, color recognition, and number classification are then used to perform fine-grained recognition within the vehicle image region, obtaining more vehicle information, thereby accelerating processing speed and reducing computational power requirements.

[0091] In a preferred embodiment, step S140 includes:

[0092] S141. Obtain the coordinates of the center position of the vehicle image region in the video frame.

[0093] S142. Obtain the positioning information in the corresponding high-precision map based on the coordinates of the center position, without limitation. In other variations, the center coordinates of the vehicle's front area can also be used as the vehicle's coordinates, etc., without limitation.

[0094] The entire system of this invention can consist of an identification server and high-position cameras distributed throughout the park. The cameras are connected to the identification server via a network. Each camera reads real-time images from each camera through network protocols, and each frame is processed and analyzed using algorithms. The high-altitude, top-down view provides a wide field of view, avoiding the obstruction issues common with low-position cameras. The high-altitude perspective, through image stitching, can provide 100% coverage of the entire park. All cameras are connected to the central identification server for centralized identification and analysis, with shared computing power among the cameras. It can identify all known vehicles, regardless of size or shape. It can identify vehicles across different cameras and compare identification features to determine if they are the same vehicle, achieving cross-camera ReID functionality. Vehicle license plates can be bound to vehicle IDs, enabling license plate retrieval within the park. The entire trajectory of external vehicles within the park can be searched using license plate information. Identification is performed using existing cameras within the park, eliminating the need for additional vehicle identification or positioning hardware. External vehicles are identified seamlessly, without requiring drivers to install or bind any equipment.

[0095] To reduce the number of surveillance cameras and monitor a wider area, it is not necessary for all areas to be covered by cameras. Instead, cameras can be set up in certain areas, and blind spots between cameras can be filled by algorithms.

[0096] In a preferred embodiment, step S150 includes:

[0097] S151. When the local maps of the sensors do not overlap, based on the positional relationship of the corresponding local maps of the sensors, the driving trajectory of the target vehicle in the corresponding local map of a sensor, and the driving speed, the prediction time of reaching the local map of the next adjacent sensor and the prediction region of the video frame that first enters the adjacent sensor are predicted along the driving trajectory direction. For example, the trajectory position can be predicted by Kalman filtering based on the time-series driving trajectory of the vehicle in the local map, and the target trajectory queue can be predicted using Kalman filtering to generate trajectory prediction. Furthermore, since the positional relationship of the local maps corresponding to each sensor is known, it is possible to predict from which side of the image of which sensor the vehicle will enter the frame (e.g., the left side of the frame, the right side of the frame, the top side of the frame, the bottom side of the frame, etc.) according to the trajectory. Subsequently, image recognition (convolution algorithm, etc.) can be performed on only one local area, thereby reducing the amount of computation.

[0098] S152. Based on the prediction time, the prediction area of ​​the image in the video frame is identified. When only one vehicle is identified in the prediction area and the time difference between the arrival time of the vehicle and the prediction time is less than a preset threshold, the motion trajectories under the two sensors are merged as the motion trajectory of the target vehicle. Finally, the motion trajectory of the missing area between the two is made up according to the vehicle's driving trajectory in the local map corresponding to the previous sensor and the driving trajectory in the local map corresponding to the current sensor, and the motion trajectory is composed together to form a complete motion trajectory, which is not limited to this.

[0099] In some cases, two vehicles may enter the frame simultaneously at the preset arrival time, requiring an additional judgment algorithm to determine which one is the target vehicle. In a preferred embodiment, step S150 includes:

[0100] S153. When the local maps of the sensors do not overlap, based on the positional relationship of the local maps corresponding to the sensors, the driving trajectory and speed of the target vehicle in the local map corresponding to a sensor, the prediction time of reaching the local map of the next adjacent sensor along the driving trajectory direction and the prediction area of ​​the video frame that preferentially enters the adjacent sensor are predicted.

[0101] S154. Based on the prediction time, the prediction region of the image in the video frame is identified. When only a few vehicles are identified in the prediction region, the existing image algorithm is used to compare the graphic differences between the vehicle trajectory and the driving trajectory. The vehicle with the smallest graphic difference is taken as the target vehicle. The motion trajectories of the target vehicles are merged, and the motion trajectories of the missing areas between the motion trajectories are supplemented, but this is not limited to this.

[0102] In a preferred embodiment, step S150 includes:

[0103] S155. When the local maps of the sensors do not overlap, based on the positional relationship of the local maps corresponding to the sensors, the driving trajectory and speed of the target vehicle in the local map corresponding to a sensor, the prediction time of reaching the local map of the next adjacent sensor along the driving trajectory direction and the prediction area of ​​the video frame that preferentially enters the adjacent sensor.

[0104] S156. Based on the prediction time, identify the prediction region of the image in the video frame. When only a few vehicles are identified in the prediction region, the similarity of the license plate information and / or attribute information of each vehicle is weighted and summed (it can be an existing weighting algorithm, but is not limited to it) to obtain a similarity reference value. The vehicle with the largest similarity reference value is taken as the target vehicle. The motion trajectories of the target vehicles are merged, and the motion trajectories of the missing areas between the motion trajectories are supplemented, but is not limited to it.

[0105] Therefore, this invention can accurately and continuously track the trajectories of multiple vehicles in a semi-enclosed environment using a small number of cameras, and greatly reduces the required hardware costs and computing power.

[0106] Figures 2 to 6 This is a schematic diagram illustrating the process of implementing the vehicle tracking method based on a semi-enclosed scenario according to the present invention. Figure 2 , 3 As shown, firstly, several surveillance cameras 11 and 12, etc., are set up in a semi-enclosed scene to capture and transmit video streams. Surveillance cameras 11 and 12 are fixed-angle high-position cameras, 40 meters above the ground. By acquiring overhead video from a top-down perspective, they can simultaneously capture vehicle features and surrounding environmental features. The local maps of surveillance cameras 11 and 12 do not overlap. Based on the overhead video intervals, video frames from surveillance cameras 11 and 12 are extracted and compared with various local areas of a high-precision map (not shown in the figure). Image similarity calculations are performed, and the local map with the highest image similarity is taken as the monitoring range corresponding to the overhead video. In this embodiment, the local map corresponding to surveillance camera 11 is the area of ​​intersection 12A, and the local map corresponding to surveillance camera 12 is the area of ​​intersection 12A, thus establishing a mapping relationship between the pixels representing the ground in the video frame image and the positioning information in the high-precision map. The image comparison algorithm using existing technology will not be elaborated here.

[0107] At this moment, truck 21 has just entered the area of ​​intersection 12A, and car 22 is about to enter the area of ​​intersection 12A. A trained first neural network performs image recognition on the video frames of the overhead video captured by monitoring probe 11, obtaining a vehicle image area (corresponding to truck 21) in the overhead video. A trained second neural network identifies the vehicle image area (corresponding to truck 21) to obtain vehicle information, including at least the license plate information A123456. The vehicle information also includes attribute information, such as body color and vehicle type; the truck's body color is orange, and its vehicle type is truck. The coordinates of the center position of the vehicle image area (corresponding to truck 21) in the video frame are obtained, and the corresponding high-precision map positioning information is obtained based on the center position coordinates. Using the positioning of truck 21 obtained from each video frame of monitoring probe 11, and the temporal continuity between video frames, the trajectory 21A of truck 21 within the area of ​​intersection 12A is established. Similarly, after the car enters the area of ​​intersection 12A, two objects can be tracked simultaneously from the overhead video captured by monitoring camera 11, establishing the trajectory 22A of car 22 within the area of ​​intersection 12A. The car's license plate information is B654321, its body color is white, and its vehicle type is sedan. Based on the positional relationship of the corresponding local maps of the sensors, the target vehicle's trajectory in the local map corresponding to monitoring camera 11, and its speed, the predicted trajectory to reach the local map of monitoring camera 12 is predicted along the direction of the trajectory. Specifically, the trajectory 21A based on truck 21 uses an existing trajectory prediction algorithm to obtain the time-series-based predicted trajectory 21C of truck 21. Similarly, the trajectory 22A based on car 22 uses an existing trajectory prediction algorithm (e.g., Kalman filtering prediction algorithm) to obtain the time-series-based predicted trajectory 22C of car 22. (For example, the trajectory position can be predicted by Kalman filtering based on the time-series driving trajectory of a vehicle within a local map, and the target trajectory queue can be predicted using Kalman filtering to generate trajectory prediction.)

[0108] like Figure 4As shown, since the positional relationships of the local maps corresponding to each sensor are known, it is possible to predict from which side of the image of which sensor the vehicle will enter the scene based on its trajectory (e.g., the left side, right side, top, or bottom of the image). Based on predicted trajectories 21C and 22C, the predicted time when the truck 21 and the car 22 reach the local map corresponding to the monitoring probe 12 and the predicted image area (upper part of the image) of the video frame of the sensor that first enters the monitoring probe 12 can be obtained, respectively. Subsequently, image recognition (convolution algorithm, etc.) can be performed on only one part, thereby reducing the amount of computation. Based on the predicted time, recognition is performed in the predicted area of ​​the video frame image. When only a few vehicles are recognized in the predicted area, the monitoring probe 12 continues to track the vehicle trajectory, obtaining trajectories 21B and 22B. The existing image algorithm is used to predict the graphic differences between trajectory 21C and trajectories 21B and 22B respectively. The trajectory with the smallest graphic difference is taken as the trajectory of truck 21. Trajectories 21A and 21B of truck 21 are merged. Based on trajectories 21A and 21B, the time-series-based trajectory 21D is completed to fill the gap between trajectories 21A and 21B. Then the complete trajectory of truck 21 passing through the area corresponding to monitoring probes 11 and 12 is trajectory 21A + trajectory 21D + trajectory 21B (if it is to save computing power, the complete trajectory can also be trajectory 21A + trajectory 21C + trajectory 21B, and trajectory 21D does not need to be calculated). Similarly, by using existing image algorithms to predict the graphic differences between trajectory 22C and trajectories 21B and 22B respectively, the trajectory with the smallest graphic difference is taken as the trajectory of car 22. The trajectories 22A and 22B of car 22 are merged, and the missing trajectory 22D between trajectories 22A and 22B is completed based on trajectories 22A and 22B. Then the complete trajectory of car 22 passing through the areas corresponding to monitoring probes 11 and 12 is trajectory 22A + trajectory 22D + trajectory 22B, which will not be elaborated here.

[0109] In a variation, when only a few vehicles are identified in the prediction area, the similarity of the license plate information and / or attribute information (including body color and vehicle type) of each vehicle is weighted and summed (this can be an existing weighting algorithm, but is not limited to it) to obtain a similarity reference value. The vehicle with the largest similarity reference value is taken as the target vehicle, the motion trajectories of the target vehicles are merged, and the motion trajectories of the missing areas between the motion trajectories are filled in.

[0110] Therefore, this invention can accurately and continuously track the trajectories of multiple vehicles in a semi-enclosed environment using a small number of cameras. Even if the monitoring ranges of the cameras do not overlap, the trajectories can be merged through trajectory prediction and vehicle information comparison algorithms to achieve trajectory tracking in the entire scene, and greatly reduce the required hardware costs and computing power.

[0111] Figure 7 This is a schematic diagram of the vehicle tracking system based on a semi-enclosed scenario according to the present invention. Figure 7 As shown, the vehicle tracking system 5 based on a semi-enclosed scenario of the present invention includes:

[0112] The video acquisition module 51 acquires overhead video through several sensors positioned at a high position in a semi-enclosed scene.

[0113] The monitoring partition module 52 matches the high-precision map of the semi-enclosed scene with the video frames of the overhead video to obtain the local map corresponding to the overhead video.

[0114] The image recognition module 53 performs image recognition on the video frames of the overhead view video to obtain at least one vehicle image region and corresponding vehicle information in the overhead view video.

[0115] The map positioning module 54 obtains the vehicle's position within a local map based on the position of the vehicle image area in the video frame image.

[0116] The motion trajectory module 55 predicts the trajectory of the target vehicle to the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, so as to establish a mapping relationship between vehicle information and the motion trajectory of the vehicle in a high-precision map of a semi-closed scene.

[0117] In a preferred embodiment, the video acquisition module 51 is configured to set up sensors at several high positions in a semi-enclosed scene. The height of the high positions is greater than the maximum height of the vehicle in the semi-enclosed scene, and the sensors acquire overhead video from a top-down perspective.

[0118] In a preferred embodiment, the monitoring partition module 52 is configured to perform image similarity calculations between video frames of the overhead view video and various local areas of the high-precision map, take the local map with the highest image similarity as the monitoring range corresponding to the overhead view video, and establish a mapping relationship between the pixels representing the ground in the video frame image and the positioning information in the high-precision map.

[0119] In a preferred embodiment, the image recognition module 53 is configured to perform image recognition on video frames of the overhead view video using a trained first neural network to obtain at least one vehicle image region in the overhead view video. The vehicle image region is then identified using a trained second neural network to obtain vehicle information, which includes at least license plate information.

[0120] In a preferred embodiment, the vehicle information further includes attribute information, which includes at least one of the following: vehicle body color, vehicle type, front shape, rear shape, vehicle markings, and type of vehicle-mounted work arm.

[0121] In a preferred embodiment, the map positioning module 54 is configured to obtain the coordinates of the center position of the vehicle image area in the image of the video frame, and obtain the positioning information in the corresponding high-precision map based on the coordinates of the center position.

[0122] In a preferred embodiment, the motion trajectory module 55 is configured to, when the local maps of the sensors do not overlap, predict the predicted time of arrival at the local map of the next adjacent sensor and the predicted region of the video frame that first enters the adjacent sensor, based on the positional relationship of the corresponding local maps of the sensors, the driving trajectory of the target vehicle in the corresponding local map of a sensor, and the driving speed, along the driving trajectory direction. Identification is performed based on the predicted region of the video frame image according to the predicted time. If only one vehicle is identified in the predicted region, and the time difference between the vehicle's arrival time and the predicted time is less than a preset threshold, then the motion trajectories under the two sensors are merged as the motion trajectory of the target vehicle, and the missing motion trajectory between the motion trajectories is supplemented.

[0123] In a preferred embodiment, the motion trajectory module 55 is configured to, when the local maps of the sensors do not overlap, predict the predicted time of reaching the local map of the next adjacent sensor and the predicted region of the video frame image that first enters the adjacent sensor, based on the positional relationship of the corresponding local maps of the sensors, the driving trajectory of the target vehicle in the corresponding local map of a sensor, and the driving speed, along the driving trajectory direction. Identification is performed in the predicted region of the video frame image based on the predicted time. If only several vehicles are identified in the predicted region, the graphic difference between the vehicle's trajectory and the driving trajectory is compared. The vehicle with the smallest graphic difference is selected as the target vehicle, the motion trajectories of the target vehicles are merged, and the missing motion trajectories between the motion trajectories are supplemented.

[0124] In a preferred embodiment, the motion trajectory module 55 is configured to, when the local maps of the sensors do not overlap, predict the predicted time of reaching the local map of the next adjacent sensor and the predicted region of the video frame that first enters the adjacent sensor, based on the positional relationship of the corresponding local maps of the sensors, the driving trajectory of the target vehicle in the corresponding local map of a sensor, and the driving speed, along the driving trajectory direction. Identification is performed in the predicted region of the video frame image based on the predicted time. When only several vehicles are identified in the predicted region, a similarity reference value is obtained by weighted summation of the similarity of the license plate information and / or various attribute information of each vehicle. The vehicle with the largest similarity reference value is selected as the target vehicle, the motion trajectories of the target vehicles are merged, and the motion trajectories of missing areas between the motion trajectories are supplemented.

[0125] The vehicle tracking system based on a semi-enclosed scenario of the present invention can accurately and continuously track the trajectories of multiple vehicles in a semi-enclosed scenario using a small number of cameras, and greatly reduces the required hardware costs and computing power.

[0126] This invention also provides a vehicle tracking device based on a semi-enclosed scenario, including a processor and a memory storing executable instructions for the processor. The processor is configured to execute steps of a vehicle tracking method based on a semi-enclosed scenario by executing the executable instructions.

[0127] As described above, the vehicle tracking device based on a semi-enclosed scenario of the present invention can accurately and continuously track the trajectories of multiple vehicles in a semi-enclosed scenario using a small number of cameras, and greatly reduces the required hardware costs and computing power.

[0128] Those skilled in the art will understand that various aspects of the present invention can be implemented as systems, methods, or program products. Therefore, various aspects of the present invention can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "platform."

[0129] Figure 8 This is a schematic diagram of the vehicle tracking device based on a semi-enclosed scenario according to the present invention. See below for reference. Figure 8 To describe an electronic device 600 according to this embodiment of the present invention. Figure 8 The electronic device 600 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0130] like Figure 8 As shown, the electronic device 600 is presented in the form of a general-purpose computing device. The components of the electronic device 600 may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including storage unit 620 and processing unit 610), a display unit 640, etc.

[0131] The storage unit stores program code, which can be executed by the processing unit 610 to perform the steps described in the above-described section on the electronic prescription transfer processing method according to various exemplary embodiments of the present invention. For example, the processing unit 610 can perform actions such as... Figure 1 The steps are shown in the figure.

[0132] Storage unit 620 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 6201 and / or cache memory 6202, and may further include a read-only memory (ROM) 6203.

[0133] Storage unit 620 may also include a program / utility 6204 having a set (at least one) program module 6205, such program module 6205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0134] Bus 630 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0135] Electronic device 600 can also communicate with one or more external devices 700 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 600, and / or with any device that enables electronic device 600 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 650. Furthermore, electronic device 600 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 660. Network adapter 660 can communicate with other modules of electronic device 600 via bus 630. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms.

[0136] This invention also provides a computer-readable storage medium for storing a program that, when executed, implements the steps of a vehicle tracking method based on a semi-closed scenario. In some possible implementations, various aspects of the invention can also be implemented as a program product comprising program code that, when run on a terminal device, causes the terminal device to perform the steps described in the above-described electronic prescription processing method section of this specification according to various exemplary embodiments of the invention.

[0137] As shown above, when the program of the computer-readable storage medium of this embodiment is executed, it can accurately and continuously track the trajectories of multiple vehicles in a semi-enclosed scene using a small number of cameras, and greatly reduces the required hardware cost and computing power.

[0138] Figure 9 This is a schematic diagram of the structure of the computer-readable storage medium of the present invention. (Reference) Figure 9As shown, a program product 800 for implementing the above-described method according to an embodiment of the present invention is described. This product may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, the readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0139] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0140] Computer-readable storage media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0141] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0142] In summary, the vehicle tracking method, system, device, and storage medium based on semi-enclosed scenarios of the present invention can accurately and continuously track the trajectories of multiple vehicles in a semi-enclosed scenario using a small number of cameras, while greatly reducing the required hardware costs and computing power.

[0143] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A vehicle tracking method based on a semi-enclosed scenario, characterized in that, Includes the following steps: The overhead video is collected by several sensors set at a high position in a semi-enclosed scene, where the high position is greater than 10 meters above the ground; The high-precision map of the semi-enclosed scene is matched with the video frames of the overhead video to obtain a local map corresponding to the overhead video. Image recognition is performed on the video frames of the overhead view video to obtain at least one vehicle image region and corresponding vehicle information in the overhead view video; The position of the vehicle within the local map is obtained based on the position of the vehicle image region in the image of the video frame; as well as Based on the trajectory of the target vehicle in the local map corresponding to one sensor, the trajectory to the next sensor is predicted, so as to establish a mapping relationship between the vehicle information and the vehicle's motion trajectory in the high-precision map of the semi-enclosed scene.

2. The vehicle tracking method based on a semi-enclosed scenario according to claim 1, characterized in that, The method of acquiring overhead video through several sensors positioned at a high vantage point in a semi-enclosed environment includes: The sensor is installed at several high-positioned cameras within the semi-enclosed environment, the height of which exceeds the maximum height of vehicles allowed to travel within the semi-enclosed environment; and The sensor acquires overhead video from a top-down perspective.

3. The vehicle tracking method based on a semi-enclosed scenario according to claim 1, characterized in that, The process of matching the high-precision map based on the semi-enclosed scene with the video frames of the overhead view video to obtain a local map corresponding to the overhead view video includes: The sensor has a fixed viewing angle and performs image similarity calculations based on the video frames of the overhead video and various parts of the high-precision map. The local map with the highest image similarity is used as the monitoring range corresponding to the overhead video; and Establish a mapping relationship between the pixels representing the ground in the video frame and the positioning information in the high-precision map.

4. The vehicle tracking method based on a semi-enclosed scenario according to claim 1, characterized in that, The process of matching the high-precision map based on the semi-enclosed scene with the video frames of the overhead view video to obtain a local map corresponding to the overhead view video includes: The sensor has a rotatable viewing angle. After each rotation, image similarity calculation is performed between the video frames of the top-down video and various parts of the high-precision map. The local map with the highest image similarity is used as the monitoring range corresponding to the overhead video; and Establish a mapping relationship between the pixels representing the ground in the video frame and the positioning information in the high-precision map.

5. The vehicle tracking method based on a semi-enclosed scenario according to claim 1, characterized in that, The step of performing image recognition on the video frames of the overhead view video to obtain at least one vehicle image region and corresponding vehicle information in the overhead view video includes: Image recognition is performed on video frames of the overhead view video using a trained first neural network to obtain at least one vehicle image region in the overhead view video; and The vehicle image region is identified by a trained second neural network to obtain vehicle information, which includes at least license plate information.

6. The vehicle tracking method based on a semi-enclosed scenario according to claim 1, characterized in that, The step of obtaining the vehicle's position within the local map based on the position of the vehicle image region in the video frame includes: Obtain the coordinates of the center position of the vehicle image region in the image of the video frame; and Based on the coordinates of the central location, the corresponding positioning information in the high-precision map is obtained.

7. The vehicle tracking method based on a semi-enclosed scenario according to claim 1, characterized in that, The method of predicting the trajectory of the target vehicle to the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, in order to establish a mapping relationship between the vehicle information and the vehicle's motion trajectory in the high-precision map of the semi-enclosed scene, includes: When the local maps of the sensors do not overlap, based on the positional relationship between the sensors and their corresponding local maps, the target vehicle's trajectory and speed within a local map corresponding to a sensor, the prediction time for reaching the local map of the next adjacent sensor and the prediction region of the image that preferentially enters the video frame of that adjacent sensor are predicted along the direction of the trajectory; and Based on the predicted time, the image in the video frame is identified. When only one vehicle is identified in the predicted area, and the time difference between the arrival time of the vehicle and the predicted time is less than a preset threshold, the motion trajectories under the two sensors are merged as the motion trajectory of the target vehicle. The motion trajectory of the missing area between the motion trajectories is supplemented based on the motion trajectories of the target vehicle under different sensors.

8. The vehicle tracking method based on a semi-enclosed scenario according to claim 1, characterized in that, The method of predicting the trajectory of the target vehicle to the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, in order to establish a mapping relationship between the vehicle information and the vehicle's motion trajectory in the high-precision map of the semi-enclosed scene, includes: When the local maps of the sensors do not overlap, based on the positional relationship between the sensors and their corresponding local maps, the driving trajectory and speed of the target vehicle in the local map corresponding to one sensor, the prediction time for reaching the local map of the next adjacent sensor is predicted along the direction of the driving trajectory, and the prediction region of the local image of the video frame that preferentially enters the adjacent sensor; and Based on the prediction time, identification is performed in the prediction region of the image in the video frame. When only a few vehicles are identified in the prediction region, the graphic difference between the trajectory of the vehicle and the driving trajectory is compared. The vehicle with the smallest graphic difference is taken as the target vehicle. The motion trajectories of the target vehicles are merged, and the motion trajectories of the target vehicles in the missing areas between the motion trajectories are supplemented based on the motion trajectories of the target vehicles under different sensors.

9. The vehicle tracking method based on a semi-enclosed scenario according to claim 5, characterized in that, The method of predicting the trajectory of the target vehicle to the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, in order to establish a mapping relationship between the vehicle information and the vehicle's motion trajectory in the high-precision map of the semi-enclosed scene, includes: When the local maps of the sensors do not overlap, based on the positional relationship between the sensors and their corresponding local maps, the target vehicle's trajectory and speed within a local map corresponding to a sensor, the prediction time for reaching the local map of the next adjacent sensor and the prediction region of the image that preferentially enters the video frame of that adjacent sensor are predicted along the direction of the trajectory; and Based on the predicted time, identification is performed in the prediction region of the image in the video frame. When only a few vehicles are identified in the prediction region, the similarity of the license plate information and / or attribute information of each vehicle is weighted and summed to obtain a similarity reference value. The vehicle with the largest similarity reference value is taken as the target vehicle. The motion trajectories of the target vehicles are merged, and the motion trajectory of the missing area between the motion trajectories is supplemented based on the motion trajectory of the target vehicle under different sensors.

10. A vehicle tracking system based on a semi-enclosed scenario, characterized in that, The system includes: The video acquisition module acquires overhead video by using several sensors positioned at a high camera position in a semi-enclosed scene, where the high camera position is more than 10 meters above the ground. The monitoring zoning module matches the high-precision map of the semi-enclosed scene with the video frames of the overhead video to obtain a local map corresponding to the overhead video. The image recognition module performs image recognition on the video frames of the overhead view video to obtain at least one vehicle image region and corresponding vehicle information in the overhead view video; The map positioning module obtains the vehicle's position within the local map based on the position of the vehicle image region within the video frame; and The motion trajectory module predicts the trajectory of the target vehicle to the next sensor based on the trajectory of the target vehicle in the local map corresponding to one sensor, so as to establish a mapping relationship between the vehicle information and the motion trajectory of the vehicle in the high-precision map of the semi-enclosed scene.

11. A vehicle tracking device based on a semi-enclosed scenario, characterized in that, include: processor; A memory in which executable instructions of the processor are stored; The processor is configured to execute the steps of the vehicle tracking method based on any one of claims 1 to 9 via executing the executable instructions.

12. A computer-readable storage medium for storing a program, characterized in that, When the program is executed, it implements the steps of the vehicle tracking method based on a semi-closed scenario as described in any one of claims 1 to 9.

Images