Vehicle real-time tracking method and device, computer device and storage medium

Abstract

The application relates to the technical field of vehicle tracking, in particular to a vehicle real-time tracking method and device, computer equipment and a storage medium, wherein the method comprises the following steps: acquiring image information of a target tracking vehicle collected by multiple roadside cameras; based on the relative relationship among the multiple roadside cameras and a pre-constructed space-time-appearance model, obtaining a passing time of the target tracking vehicle passing through a blind area among the multiple roadside cameras based on the image information; and obtaining traffic behavior characteristics of the target tracking vehicle by using the passing time and a preset normal distribution model, and generating a vehicle real-time tracking result based on the traffic behavior characteristics. Thus, the problems in the prior art that, due to the existence of a blind area among the visual fields of multiple cameras and the diversification of the target motion mode, the appearance and disappearance time of the target cannot be accurately predicted, the accuracy of tracking is affected, the difficulty of information acquisition is relatively high, the complexity of calculation is relatively high, the calculation rate is reduced, and the real-time performance of the algorithm is affected are solved.

Description

Technical Field

[0001] This application relates to the field of vehicle tracking technology, and in particular to a method, apparatus, computer equipment, and storage medium for real-time vehicle tracking. Background Technology

[0002] In related technologies, current cross-camera vehicle target tracking techniques are mainly implemented through two methods: target appearance feature models and spatiotemporal models. Among them, vehicle target tracking methods based on target appearance feature models mainly acquire color, geometric, and texture features for associated tracking. Since color features are easy to acquire and have strong stability when the target pose changes, many researchers have focused on using the target's color features to build an appearance model for cross-camera target matching. In the early 20th century, Porikli et al. proposed the concept of a brightness transfer function, which solves the problem of large changes in target brightness during cross-camera tracking by reasonably constructing brightness transfer functions between different cameras. In addition to using color features, Collinst et al. attempted to use 3D scene models between multiple cameras to assist cross-camera target matching. They also utilized the rotation invariance and scale scaling invariance properties of the SIFT (Scale Invariant Feature Transform) operator to use region SIFT features as an appearance model for multi-camera target tracking. The multi-camera vehicle target tracking method based on the spatiotemporal model mainly projects the target onto the world coordinate system based on the relationship between the topological structures of the field of view of the cameras. By using spatial information to calculate trajectory similarity, it can be further associated with the actual moving target, thereby achieving the purpose of tracking.

[0003] However, in related technologies, due to blind spots between the fields of view of multiple cameras, the diverse movement patterns of targets, and the inability to accurately predict the appearance and disappearance times of targets, as well as inter-class differences in target feature matching, target association fails, affecting the accuracy of tracking. It is necessary to obtain the intrinsic and extrinsic parameters of each camera, which is difficult to obtain and has a high degree of computational complexity, reducing the computational speed and affecting the real-time performance of the algorithm, thus affecting the real-time tracking of vehicles. Improvements are urgently needed. Summary of the Invention

[0004] This application provides a vehicle real-time tracking method, apparatus, computer equipment, and storage medium to solve the problems in related technologies, such as the existence of blind spots between the fields of view of multiple cameras, the diversity of target movement patterns, the inability to accurately predict the appearance and disappearance time of the target, which affects the accuracy of tracking, the difficulty in information acquisition, the high complexity of calculation, the reduction of calculation speed, and the impact on the real-time performance of the algorithm.

[0005] The first aspect of this application provides a real-time vehicle tracking method, comprising the following steps: acquiring image information of a target vehicle collected by multiple roadside cameras; based on the relative relationship between the multiple roadside cameras and a pre-constructed space-time-appearance model, obtaining the travel time of the target vehicle traversing the blind spots between the multiple roadside cameras based on the image information; and using the travel time and a preset normal distribution model to obtain traffic behavior characteristics of the target vehicle, and generating a real-time vehicle tracking result based on the traffic behavior characteristics.

[0006] Optionally, in one embodiment of this application, the distance between the plurality of roadside cameras is less than a preset distance and they are arranged in a mirror image on the roadside.

[0007] Optionally, in one embodiment of this application, the step of obtaining the travel time of the target tracking vehicle in the blind spot between the multiple roadside cameras based on the image information, based on the relative relationship between the multiple roadside cameras and a pre-constructed space-time-appearance model, includes: obtaining the pixel coordinates of the moment when the target tracking vehicle disappears from the image information of any roadside camera; when the target tracking vehicle detects a new target in the image information of an adjacent roadside camera of any roadside camera, searching for the pixel coordinates of the moment of disappearance from the pixel coordinates of the new target; and searching for and associating with the vehicle closest to the pixel coordinates of the new target from the vicinity of the pixel coordinates of the moment of disappearance.

[0008] Optionally, in one embodiment of this application, the step of obtaining the traffic behavior characteristics of the target tracking vehicle using the passage time and a preset normal distribution model, and generating real-time vehicle tracking results based on the traffic behavior characteristics, includes: obtaining the time difference between the new target and vehicles within the corresponding search range of the adjacent roadside cameras; determining whether the time difference satisfies a preset confidence interval; if the preset confidence interval is satisfied, then using a preset time model to associate the target, otherwise using a preset appearance feature model to associate the target and generate association information; and using the association information to update the normal distribution model to obtain an updated normal distribution model.

[0009] Optionally, in one embodiment of this application, the step of obtaining the traffic behavior characteristics of the target tracking vehicle using the passage time and a preset normal distribution model, and generating a real-time vehicle tracking result based on the traffic behavior characteristics, further includes: obtaining the real-time vehicle tracking result based on the updated normal distribution model.

[0010] A second aspect of this application provides a real-time vehicle tracking device, comprising: an acquisition module for acquiring image information of a target vehicle collected by multiple roadside cameras; a prediction module for obtaining the travel time of the target vehicle through the blind spots between the multiple roadside cameras based on the relative relationship between the multiple roadside cameras and a pre-constructed space-time-appearance model; and a generation module for acquiring traffic behavior characteristics of the target vehicle using the travel time and a preset normal distribution model, and generating a real-time vehicle tracking result based on the traffic behavior characteristics.

[0011] Optionally, in one embodiment of this application, the distance between the plurality of roadside cameras is less than a preset distance and they are arranged in a mirror image on the roadside.

[0012] Optionally, in one embodiment of this application, the prediction module includes: a first acquisition unit, configured to acquire the pixel coordinates of the moment when the target tracking vehicle disappears from the image information of any roadside camera; a search unit, configured to, when the target tracking vehicle detects a new target in the image information of an adjacent roadside camera of any roadside camera, search for the pixel coordinates of the moment of disappearance from the pixel coordinates of the new target; and an association unit, configured to search for and associate with the vehicle closest to the pixel coordinates of the new target from the vicinity of the pixel coordinates of the moment of disappearance.

[0013] Optionally, in one embodiment of this application, the generation module includes: a second acquisition unit, configured to acquire the time difference between the new target and vehicles within the corresponding search range of the adjacent roadside camera; a judgment unit, configured to judge whether the time difference satisfies a preset information interval; a generation unit, configured to perform target association using a preset time model when the preset information interval is satisfied, otherwise perform target association using a preset appearance feature model, and generate association information; and an update unit, configured to update the normal distribution model using the association information to obtain an updated normal distribution model.

[0014] Optionally, in one embodiment of this application, it further includes: a target tracking module, used to obtain the real-time tracking result of the vehicle based on the updated normal distribution model.

[0015] A third aspect of this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the real-time vehicle tracking method as described in the above embodiments.

[0016] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described real-time vehicle tracking method.

[0017] This application embodiment can determine the travel time of a target vehicle through blind spots between multiple roadside cameras based on image information collected by multiple roadside cameras. It then combines this with a preset normal distribution model to obtain the traffic behavior characteristics of the target vehicle and generate real-time vehicle tracking results. This achieves self-iterative optimization of the normal distribution model, improving the overall accuracy of the multi-target association method, reducing the association error rate, and effectively reducing the complexity of the multi-target association algorithm, increasing the computational speed, and ensuring the real-time performance of the algorithm. Therefore, it solves the problems in related technologies, such as the inability to accurately predict the appearance and disappearance time of targets due to blind spots between the fields of view of multiple cameras and the diverse movement patterns of targets, which affects the accuracy of tracking, and the high difficulty in information acquisition and computational complexity, which reduces the computational speed and affects the real-time performance of the algorithm.

[0018] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0019] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0020] Figure 1 This is a flowchart of a real-time vehicle tracking method provided according to an embodiment of this application;

[0021] Figure 2 This is a schematic diagram illustrating the principle of a multi-target association method based on spatial point matching according to an embodiment of this application;

[0022] Figure 3 This is a schematic diagram of the normal distribution of the time when traffic vehicles pass through the blind spot of a 15-meter camera on the same speed-limited road segment according to an embodiment of this application;

[0023] Figure 4 This is a schematic diagram illustrating the principle of a multi-objective association method based on a spatial-temporal-appearance feature model according to an embodiment of this application;

[0024] Figure 5 This is a schematic diagram of the structure of a real-time vehicle tracking device according to an embodiment of this application;

[0025] Figure 6 This is a schematic diagram of the structure of a computer device provided according to an embodiment of this application. Detailed Implementation

[0026] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0027] The following description, with reference to the accompanying drawings, outlines a vehicle real-time tracking method, apparatus, computer device, and storage medium according to embodiments of this application. Addressing the issues raised in the background section regarding related technologies, such as blind spots between multiple camera viewpoints, diverse target movement patterns, the inability to accurately predict target appearance and disappearance times, impacting tracking accuracy, difficulty in information acquisition, high computational complexity, reduced computational speed, and affecting algorithm real-time performance, this application provides a vehicle real-time tracking method. This method determines the travel time of a target vehicle through blind spots between multiple roadside cameras based on image information collected by multiple roadside cameras. It then combines this with a preset normal distribution model to obtain the traffic behavior characteristics of the target vehicle and generate real-time vehicle tracking results. This achieves self-iterative optimization of the normal distribution model, improving the overall accuracy of the multi-target association method, reducing the association error rate, effectively reducing the complexity of the multi-target association algorithm, increasing computational speed, and ensuring the algorithm's real-time performance. This solves the problems in related technologies, such as blind spots between multiple camera fields of view, diverse target movement patterns, inability to accurately predict the appearance and disappearance time of targets, affecting tracking accuracy, difficulty in information acquisition, high computational complexity, reduced computation speed, and impact on algorithm real-time performance.

[0028] Specifically, Figure 1 This is a flowchart illustrating a real-time vehicle tracking method provided in an embodiment of this application.

[0029] like Figure 1 As shown, the real-time vehicle tracking method includes the following steps:

[0030] In step S101, image information of the target tracking vehicle is acquired from multiple roadside cameras.

[0031] In some embodiments, the embodiments of this application can analyze and extract images using computer vision and image processing techniques to obtain image information of the target vehicle collected by the roadside camera. This information can be the vehicle's location information or its appearance features. For example, by analyzing the image collected by the roadside camera using computer vision and image processing techniques, the pixel coordinates of the vehicle can be obtained as (100, 200), and the vehicle's color can be black.

[0032] In the embodiments of this application, there are many ways to acquire image information, and no specific limitations are made here. The aim is to provide a comprehensive view from multiple perspectives, improve the accuracy of image information, and provide an image data foundation for subsequent target detection and tracking.

[0033] Optionally, in one embodiment of this application, the distance between multiple roadside cameras is less than a preset distance and they are arranged in a mirror image on the roadside.

[0034] Understandably, the preset distance can have different values, such as 30 meters or 50 meters, and its value can be related to the relationship between roadside cameras or the relationship between captured images.

[0035] In actual implementation, the embodiments of this application realize that the distance between multiple roadside cameras is less than a preset distance and they are arranged in a mirror image on the roadside. For example, according to preset requirements, the distance between roadside cameras can be set to less than 50 meters and they can be arranged in a mirror image on the roadside.

[0036] In the embodiments of this application, the preset distance can be set by those skilled in the art according to specific application needs and scenario requirements, without any specific limitations, to improve the accuracy and convenience of image information acquisition, so as to meet the needs of target identification and tracking.

[0037] In step S102, based on the relative relationships between multiple roadside cameras and a pre-built space-time-appearance model, the travel time of the target tracking vehicle in the blind spot between multiple roadside cameras is obtained based on image information.

[0038] It is understandable that relative relationship refers to the relative position and orientation relationship between roadside cameras. For example, the distance between roadside cameras is less than 50 meters and they can be distributed in a mirror image. The pre-built space-time-appearance model refers to the relationship between the space, time and appearance of a vehicle passing through the blind spot.

[0039] As one possible approach, embodiments of this application can predict the travel time of a target vehicle in a blind spot area based on the relative relationships between multiple roadside cameras and a pre-built space-time-appearance model, and by combining the appearance features and motion trajectory of the target vehicle obtained from image information. For example, when the target vehicle is detected entering the blind spot in the first camera, based on the vehicle's speed (e.g., 40 km / h) and direction of travel (e.g., to the right), and based on the built model and the relative relationships between the roadside cameras, the travel time of the vehicle in the blind spot can be predicted to be 1 second.

[0040] The embodiments of this application can predict the passage time of blind spots based on image information, the relative relationship between cameras and a pre-built space-time-appearance model. The algorithm has a simple structure, consumes little computing resources, and effectively improves the accuracy of the prediction results.

[0041] Optionally, in one embodiment of this application, based on the relative relationship between multiple roadside cameras and a pre-built space-time-appearance model, the travel time of the target tracking vehicle in the blind spot between multiple roadside cameras is obtained based on image information, including: obtaining the pixel coordinates of the moment when the target tracking vehicle disappears in the image information of any roadside camera; when the target tracking vehicle detects a new target in the image information of an adjacent roadside camera of any roadside camera, finding the pixel coordinates of the moment of disappearance from the pixel coordinates of the new target; and searching for the vehicle closest to the pixel coordinates of the new target in the vicinity of the pixel coordinates of the moment of disappearance and associating it with the vehicle.

[0042] In some embodiments, this application can reduce the distance between roadside cameras and arrange them in a mirror image on the roadside, thereby minimizing the blind spots of the cameras and establishing a relatively strict spatial correspondence between the fields of view of the two cameras. This results in a shorter travel distance for the target vehicle in the blind spots, making lane changes difficult. The position of the target vehicle within the camera's field of view remains relatively stable. Therefore, the corresponding position in adjacent cameras can be found based on the location where the target vehicle disappears from the forward-looking camera, and the nearest nearby vehicle can be searched for and associated with it. Figure 2 As shown, when two cameras are arranged in a mirror image and the blind spot between the cameras is less than 15 meters, whenever a vehicle leaves camera A, the pixel coordinates of the moment the vehicle disappears are recorded in real time. After a period of time, when camera B detects a new target, it needs to start from the pixel coordinates of the new target. ′ Find the corresponding pixel coordinate 1 in camera A, and search for pixel coordinates near pixel coordinate 1 that are similar to pixel coordinate 1. ′ Associate the closest vehicles.

[0043] This application embodiment can find the pixel coordinates of the disappearance time from the pixel coordinates of the new target. By utilizing the image information of adjacent cameras, visual blind spots can be compensated, ensuring the continuity of target tracking. Furthermore, it can search for and associate the known target that is closest to the pixel coordinates of the new target in the vicinity. This can more accurately find the disappearing target related to the new target, improve the accuracy of target association, and at the same time, effectively reduce the complexity of target association and improve the calculation speed.

[0044] In step S103, the traffic behavior characteristics of the target vehicle are obtained using the travel time and a preset normal distribution model, and the real-time vehicle tracking results are generated based on the traffic behavior characteristics.

[0045] It is understandable that the preset normal distribution model refers to the constructed normal distribution model of vehicle travel time to describe the probability distribution characteristics of vehicle travel time. Traffic behavior characteristics refer to information related to vehicle driving behavior and status, such as speed, passage degree, and travel path. Real-time tracking results refer to real-time information of the vehicle's current status, such as location and speed.

[0046] In actual implementation, the embodiments of this application can obtain the travel time of the target vehicle through a roadside camera, and obtain the traffic behavior characteristics of the vehicle by combining it with a preset normal distribution model, thereby generating real-time vehicle tracking results. For example, the travel time is summarized into the preset normal distribution model, the travel time t of the vehicle is calculated and the position information s of the vehicle is detected. Through data comparison and analysis, the speed v of the vehicle can be estimated, and its position s' at the next moment can be inferred, and real-time vehicle tracking can be continuously performed.

[0047] This application embodiment obtains the traffic behavior characteristics of the target vehicle being tracked, enabling quantitative analysis of vehicle driving conditions and accurate understanding of the target vehicle's driving status. By iteratively updating the target's position and status, the real-time tracking results of the vehicle can be optimized, reducing tracking errors and improving tracking accuracy and continuity.

[0048] Optionally, in one embodiment of this application, traffic behavior characteristics of the target vehicle are obtained using travel time and a preset normal distribution model, and real-time vehicle tracking results are generated based on the traffic behavior characteristics. This includes: obtaining the time difference between the new target and vehicles within the corresponding search range in adjacent roadside cameras; determining whether the time difference meets a preset information interval; if it meets the preset information interval, then a preset time model is used for target association, otherwise a preset appearance feature model is used for target association to generate association information; and updating the normal distribution model using the association information to obtain the updated normal distribution model.

[0049] Understandably, the confidence interval refers to a pre-set time difference range used to determine whether the calculated time difference is within an acceptable range, serving as the basis for selecting a time model or an appearance feature model.

[0050] In some embodiments, this application embodiment can calculate the time difference between a new target and targets in adjacent camera images, determine whether the time difference meets a preset information interval, and thus decide whether to use a time model or an appearance feature model for target association. The time model is based on the temporal correlation of targets, while the appearance feature model is based on the similarity of targets in appearance. Therefore, target association operations can be performed according to the selected model, and the generated association information can be used to update the normal distribution model, which is beneficial for subsequent target tracking and analysis.

[0051] Specifically, such as Figure 3 As shown, based on the normal distribution pattern and combined with the constructed spatial-temporal-appearance model and target association, it can be seen that when most vehicles pass through the blind spot at normal speeds, the order in which vehicles enter the blind spot is generally reliable. Therefore, a spatial-temporal model can be established as a foundation for collecting the time taken by vehicles to pass through the camera's blind spot and constructing a normal distribution model. After the normal distribution model is built, when a new target appears in the forward-looking camera, it is necessary to first determine whether the time difference between the detected target and vehicles within the corresponding search range of adjacent cameras meets the confidence interval specified by the normal distribution model. If the time difference meets the confidence interval, the temporal model is used for target association; otherwise, the appearance feature model is used. After vehicle association is completed, the normal distribution model is continuously updated using the time taken by vehicles to pass through the blind spot.

[0052] In addition, such as Figure 4 As shown, whenever a vehicle drives away from the forward-facing camera, a timestamp T is recorded for that moment. n The pixel coordinates of the vehicle in the camera (X) n ,Y n The image appearance features of the vehicle F n When the rear-view camera detects a new target, it records the timestamp T of that moment. n+1 The pixel coordinates of the vehicle in the camera (X) n+1 ,Y n+1 The image appearance features of the vehicle F n+1 The Kolmogorov-Smirnov test (KS test, goodness-of-fit test) is used to test the normal distribution model. If the collected data fails the KS test, the target needs to be associated using a space-time model, and the time difference of the vehicle's blind spot after association is continuously collected to further improve the normal distribution model. If the KS test passes, it is determined whether the time difference between the newly detected target and the vehicle in the blind spot conforms to the specified confidence interval of the normal distribution. If it conforms to the 95% confidence interval of the normal distribution, the time model is used to complete the association; otherwise, the appearance feature model is used to complete the association. In this way, the time difference of the vehicle's blind spot after association is collected to ensure the online learning of the normal distribution model.

[0053] In the embodiments of this application, the setting of the information interval can be adjusted according to specific application requirements and experience, and can be set by those skilled in the art. No specific restrictions are imposed here. A time model or appearance feature model can be reasonably selected for target association to increase the accuracy and reliability of target association, improve the comprehensiveness of considering all aspects of target association, improve the success rate of target association, and update the normal distribution model to improve the accuracy and continuity of target tracking.

[0054] Optionally, in one embodiment of this application, the traffic behavior characteristics of the target tracking vehicle are obtained by using the travel time and a preset normal distribution model, and the real-time vehicle tracking result is generated based on the traffic behavior characteristics. The method further includes obtaining the real-time vehicle tracking result based on the updated normal distribution model.

[0055] It is understood that the embodiments of this application can adjust the normal distribution according to parameters such as the target's position and speed, thereby using the updated normal distribution model to generate real-time vehicle tracking results. For example, based on the tracked vehicle's state (such as position, speed, etc.), the normal distribution parameters can be adjusted, thereby re-detecting and obtaining the vehicle's current real-time position information in the current image based on the updated normal distribution model.

[0056] The embodiments of this application, by using an updated normal distribution model, can better match the actual target behavior and state distribution, which helps to improve the accuracy of target tracking, better track the target's motion trajectory, and improve the ability to continuously track.

[0057] Combination Figure 2 , Figure 3 and Figure 4 As shown below, the real-time vehicle tracking method of this application will be described in detail with reference to an embodiment.

[0058] It is understandable that, such as Figure 2 As shown, when the blind zone between two cameras is small, the lateral distance of the target vehicle changes little when it moves from the forward-looking camera to the adjacent camera. The corresponding pixel coordinates of the camera should also change little. Therefore, the nearest vehicle in the vicinity can be searched for to complete the association. The association algorithm has a simple structure, consumes little computing resources, and can be applied to real-time cross-domain tracking deployments on real roads. Furthermore, it can effectively improve the accuracy of cross-camera tracking. Specifically, when the two cameras are mirrored and the blind zone between them is less than 15 meters, whenever a vehicle leaves camera A, the roadside unit records the pixel coordinates of the vehicle at the moment it disappears. After a period of time, when camera B detects a new target, it needs to start from the pixel coordinates of the detected target (1...). ′ Find the corresponding pixel coordinate 1 in camera A, and search for pixel coordinates near pixel coordinate 1 that are similar to pixel coordinate 1. ′ Associate the closest vehicles.

[0059] In addition, such as Figure 3As shown, the dataset used in the statistics has a sample size of 1,044,634. When vehicles pass through the blind spot between cameras at normal speeds, the order in which vehicles enter the blind spot is highly reliable. However, when vehicles pass through the blind spot at high or low speeds, the order of vehicles may change, leading to incorrect target association. Based on the normal distribution pattern, most vehicles pass through the blind spot at normal speeds, and the order in which vehicles enter the blind spot is generally reliable. After the normal distribution model is built, when a new target appears in the forward-looking camera, it is necessary to first determine whether the time difference between the target and vehicles within the corresponding search range of adjacent cameras meets the confidence interval specified by the normal distribution model. If the time difference meets the confidence interval, the time model is used for target association; otherwise, the appearance feature model is used. After vehicle association is completed, the normal distribution model is continuously updated using the vehicle passage time through the blind spot.

[0060] Specifically, such as Figure 4 As shown, whenever a vehicle drives away from the forward-facing camera, a timestamp T is recorded for that moment. n The pixel coordinates of the vehicle in the camera (X) n ,Y n The image appearance features of the vehicle F n After a period of time, when the rear-view camera detects a new target, it records the timestamp T of that moment. n+1 The pixel coordinates of the vehicle in the camera (X) n+1 ,Y n+1 The image appearance features of the vehicle F n+1 This method uses the KS test to verify the normal distribution model. If the collected data fails the KS test, the target needs to be associated using a space-time model, and the time difference of the vehicle's blind spot after association is continuously collected to further improve the normal distribution model. If the KS test passes, it is determined whether the time difference between the newly detected target and the vehicle in the blind spot conforms to the specified confidence interval of the normal distribution. If it conforms to the 95% confidence interval of the normal distribution, the association is completed using a time model; otherwise, the association is completed using an appearance feature model. After the association is completed, the time difference of the vehicle's blind spot after association also needs to be collected for online learning of the normal distribution model.

[0061] The real-time vehicle tracking method proposed in this application can determine the travel time of a target vehicle in the blind spots between multiple roadside cameras based on image information collected by multiple roadside cameras. It then combines this with a preset normal distribution model to obtain the traffic behavior characteristics of the target vehicle and generate real-time vehicle tracking results. This achieves self-iterative optimization of the normal distribution model, improving the overall accuracy of the multi-target association method, reducing the association error rate, and effectively reducing the complexity of the multi-target association algorithm, increasing the computational speed, and ensuring the real-time performance of the algorithm. Therefore, it solves the technical problems in related technologies, such as the inability to accurately predict the appearance and disappearance time of targets due to blind spots between the fields of view of multiple cameras and the diverse movement patterns of targets, which affects the accuracy of tracking, and the high difficulty in information acquisition and computational complexity, which reduces the computational speed and affects the real-time performance of the algorithm.

[0062] Next, the vehicle real-time tracking device according to the embodiments of this application is described with reference to the accompanying drawings.

[0063] Figure 5 This is a block diagram of a vehicle real-time tracking device according to an embodiment of this application.

[0064] like Figure 5 As shown, the real-time vehicle tracking device 10 includes: an acquisition module 100, a prediction module 200, and a generation module 300.

[0065] Specifically, the acquisition module 100 is used to acquire image information of the target tracking vehicle collected by multiple roadside cameras.

[0066] The prediction module 200 is used to obtain the travel time of the target vehicle in the blind spot between multiple roadside cameras based on the relative relationship between multiple roadside cameras and a pre-built space-time-appearance model, and based on image information.

[0067] The generation module 300 is used to obtain the traffic behavior characteristics of the target vehicle by utilizing the travel time and a preset normal distribution model, and to generate real-time vehicle tracking results based on the traffic behavior characteristics.

[0068] Optionally, in one embodiment of this application, the distance between multiple roadside cameras is less than a preset distance and they are arranged in a mirror image on the roadside.

[0069] Optionally, in one embodiment of this application, the prediction module 200 includes: a first acquisition unit, a search unit, and an association unit.

[0070] The first acquisition unit is used to acquire the pixel coordinates of the moment when the target tracking vehicle disappears from the image information of any roadside camera;

[0071] The searching unit is used to find the pixel coordinates of the disappearance time from the pixel coordinates of the new target when the target tracking vehicle detects a new target in the image information of the adjacent roadside camera of any roadside camera;

[0072] The association unit is used to search for the vehicle closest to the pixel coordinates of the new target from the pixel coordinates near the disappearance time and associate it with the vehicle.

[0073] Optionally, in one embodiment of this application, the generation module 300 includes: a second acquisition unit, a judgment unit, an association unit, and an update unit.

[0074] The second acquisition unit is used to acquire the time difference between the new target and the vehicles within the corresponding search range of the adjacent roadside camera;

[0075] The judgment unit is used to determine whether the time difference meets the preset information interval.

[0076] The association unit is used to perform target association using a preset time model when a preset information interval is met, and otherwise to perform target association using a preset appearance feature model to generate association information.

[0077] The update unit is used to update the normal distribution model using the correlation information to obtain the updated normal distribution model.

[0078] Optionally, in one embodiment of this application, a target tracking module is also included.

[0079] The target tracking module is used to obtain real-time vehicle tracking results based on the updated normal distribution model.

[0080] It should be noted that the foregoing explanation of the vehicle real-time tracking method embodiment also applies to the vehicle real-time tracking device of this embodiment, and will not be repeated here.

[0081] The real-time vehicle tracking device proposed in this application can determine the travel time of a target vehicle in the blind spots between multiple roadside cameras based on image information collected by multiple roadside cameras. It then combines this with a preset normal distribution model to obtain the traffic behavior characteristics of the target vehicle and generate real-time vehicle tracking results. This achieves self-iterative optimization of the normal distribution model, improving the overall accuracy of the multi-target association method, reducing the association error rate, and effectively reducing the complexity of the multi-target association algorithm, increasing the computational speed, and ensuring the real-time performance of the algorithm. Therefore, it solves the technical problems in related technologies, such as the inability to accurately predict the appearance and disappearance time of targets due to blind spots between multiple camera fields of view and the diverse movement patterns of targets, which affects the accuracy of tracking, and the high difficulty in information acquisition and computational complexity, which reduces the computational speed and affects the real-time performance of the algorithm.

[0082] Figure 6 A schematic diagram of the structure of a computer device provided in an embodiment of this application. The computer device may include:

[0083] The memory 601, the processor 602, and the computer program stored on the memory 601 and capable of running on the processor 602.

[0084] When the processor 602 executes the program, it implements the real-time vehicle tracking method provided in the above embodiments.

[0085] Furthermore, computer equipment also includes:

[0086] Communication interface 603 is used for communication between memory 601 and processor 602.

[0087] The memory 601 is used to store computer programs that can run on the processor 602.

[0088] The memory 601 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0089] If the memory 601, processor 602, and communication interface 603 are implemented independently, then the communication interface 603, memory 601, and processor 602 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0090] Optionally, in a specific implementation, if the memory 601, processor 602, and communication interface 603 are integrated on a single chip, then the memory 601, processor 602, and communication interface 603 can communicate with each other through an internal interface.

[0091] The processor 602 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0092] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described real-time vehicle tracking method.

[0093] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described 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 the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0094] Furthermore, the terms "first" and "second" are used for descriptive 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 description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0095] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0096] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0097] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0098] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0099] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0100] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for real-time vehicle tracking, characterized in that, Includes the following steps: Acquire image information of the target vehicle collected by multiple roadside cameras; Based on the relative relationships between the multiple roadside cameras and a pre-constructed space-time-appearance model, the travel time of the target tracking vehicle traversing the blind spots between the multiple roadside cameras is obtained based on the image information. This process includes: obtaining the pixel coordinates of the moment the target tracking vehicle disappears from the image information of any roadside camera; when the target tracking vehicle detects a new target in the image information of an adjacent roadside camera, finding the pixel coordinates of the moment of disappearance from the pixel coordinates of the new target; and searching for and associating with the vehicle closest to the pixel coordinates of the new target in the vicinity of the pixel coordinates of the moment of disappearance. The process involves obtaining traffic behavior characteristics of the target vehicle using the travel time and a preset normal distribution model, and generating real-time vehicle tracking results based on these characteristics. This includes: obtaining the time difference between the new target and vehicles within the corresponding search range of adjacent roadside cameras; determining whether the time difference satisfies a preset information interval; if it does, associating the target using a preset time model, otherwise associating the target using a preset appearance feature model to generate association information; updating the normal distribution model using the association information to obtain an updated normal distribution model; and obtaining the real-time vehicle tracking results based on the updated normal distribution model.

2. The method according to claim 1, characterized in that, The distance between the multiple roadside cameras is less than a preset distance, and they are arranged in a mirror image on the roadside.

3. A vehicle real-time tracking device, characterized in that, include: The acquisition module is used to acquire image information of the target tracking vehicle collected by multiple roadside cameras; The prediction module is used to obtain the travel time of the target tracking vehicle in the blind spot between the multiple roadside cameras based on the image information, based on the relative relationship between the multiple roadside cameras and the pre-built space-time-appearance model. The prediction module includes: a first acquisition unit, configured to acquire the pixel coordinates of the moment when the target tracking vehicle disappears from the image information of any roadside camera; a search unit, configured to, when the target tracking vehicle detects a new target in the image information of an adjacent roadside camera of any roadside camera, search for the pixel coordinates of the moment of disappearance from the pixel coordinates of the new target; and an association unit, configured to search for and associate with the vehicle closest to the pixel coordinates of the new target from the vicinity of the pixel coordinates of the moment of disappearance. A generation module is used to obtain traffic behavior characteristics of the target vehicle using the passage time and a preset normal distribution model, and to generate real-time vehicle tracking results based on the traffic behavior characteristics. The generation module includes: a second acquisition unit, used to acquire the time difference between the new target and vehicles within the corresponding search range of the adjacent roadside cameras; a judgment unit, used to judge whether the time difference meets a preset information interval; a generation unit, used to perform target association using a preset time model when the preset information interval is met, otherwise perform target association using a preset appearance feature model, and generate association information; an update unit, used to update the normal distribution model using the association information to obtain an updated normal distribution model; and a target tracking module, used to obtain the real-time vehicle tracking results based on the updated normal distribution model.

4. The apparatus according to claim 3, characterized in that, The distance between the multiple roadside cameras is less than a preset distance, and they are arranged in a mirror image on the roadside.

5. A computer device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the real-time vehicle tracking method as described in any one of claims 1-2.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the real-time vehicle tracking method as described in any one of claims 1-2.

Images