Vehicle trajectory restoration method, electronic device, and storage medium

Abstract

The present application discloses a vehicle trajectory restoration method, an electronic device, and a storage medium, the method comprising: obtaining the endpoint passing data of the target vehicle photographed at the entrance and exit and the point passing data of the passing vehicle photographed at a number of video points; determining the target time period for the target vehicle to pass through each video point based on the endpoint passing data and the point passing data, wherein at least one of the several video points is the first video point, and the target time period corresponding to the first video point is the first target time period; obtaining the first candidate vehicle photographed by the first video point within the first target time period; in response to the first candidate vehicle meeting the preset requirements, associating the first candidate vehicle with the target vehicle, and using the point passing data corresponding to the associated vehicle as the trajectory point passing data of the target vehicle at the first video point. The above scheme can accurately obtain the trajectory point passing data of the target vehicle on the road.

Description

Technical Field

[0001] The present application relates to the technical field of data analysis, and in particular to a vehicle trajectory restoration method, electronic equipment, and storage medium. Background Art

[0002] In the prior art, in order to determine the target vehicle's track passing data on the road, the distance and time difference between the entrance and exit gantries are usually used to calculate the target vehicle's driving speed, so as to locate the target vehicle's time and space information, and then verify it through the view information to determine the target vehicle's track point passing data. However, this solution has problems such as inaccurate time and space information and the possibility that multiple similar vehicles pass through the entrance and exit within a period of time, resulting in inaccurate track point passing data. Therefore, how to accurately obtain the target vehicle's track point passing data on the road is an urgent problem to be solved. Summary of the invention

[0003] The present application at least provides a vehicle trajectory restoration method, electronic device, and storage medium, which can accurately obtain the trajectory point passing data of the target vehicle on the road.

[0004] In a first aspect, the present application provides a vehicle trajectory restoration method, the method comprising: obtaining endpoint passing data of a target vehicle photographed at an entrance and exit and point passing data of a passing vehicle photographed at a plurality of video points, wherein the plurality of video points are arranged between the entrance and exit; determining a target time period for a target vehicle to pass through each video point based on the endpoint passing data and the point passing data, wherein at least one video point of the plurality of video points is a first video point, and the target time period corresponding to the first video point is a first target time period; obtaining a first candidate vehicle photographed at the first video point within the first target time period; in response to the first candidate vehicle meeting a preset requirement, associating the first candidate vehicle with the target vehicle, and using the point passing data corresponding to the associated vehicle as the trajectory point passing data of the target vehicle at the first video point.

[0005] Among them, the endpoint vehicle passing data includes the starting shooting time, the ending shooting time and the first distance between the entrance and exit, the starting shooting time represents the time when the target vehicle passes through the entrance of the entrance and exit, and the ending shooting time represents the time when the target vehicle passes through the exit of the entrance and exit; the point vehicle passing data includes the second distance between the entrance and the video point; based on the endpoint vehicle passing data and the point vehicle passing data, determine the target time period for the target vehicle to pass through each video point, including: obtaining the difference between the starting shooting time and the ending shooting time as the first passing time difference for the target vehicle to pass through the entrance and exit; obtaining the ratio between the first distance and the first passing time difference as the first average speed of the target vehicle; based on the first average speed and the second distance, obtain the target time period of the video point.

[0006] Among them, based on the first average speed and the second distance, the target time period of the video point is obtained, including: obtaining the ratio between the second distance and the first average speed as the predicted driving time of the target vehicle; obtaining the sum of the predicted driving time and the starting shooting time as the predicted arrival time; based on the predicted arrival time and the time difference deviation coefficient, the target time period is obtained, wherein the time difference deviation coefficient is related to the discrete degree characterization value of the first average speed of the target vehicle, the difference between the predicted arrival time and the time difference deviation coefficient is taken as the minimum value of the target time period, and the sum of the predicted arrival time and the time difference deviation coefficient is taken as the maximum value of the target time period.

[0007] Among them, obtaining the first candidate vehicle photographed at the first video point within the first target time period includes: determining the first candidate vehicle photographed within the first target time period based on the point vehicle passing data and the endpoint vehicle passing data corresponding to the first video point.

[0008] Among them, the point-by-point vehicle passing data corresponding to the first video point includes several initial captured images corresponding to each first passing vehicle, the first passing vehicle is a vehicle passing the first video point captured within the first target time period, and the endpoint-by-point vehicle passing data includes a target captured image corresponding to the entrance of the target vehicle passing through the entrance and exit; based on the point-by-point vehicle passing data and the endpoint-by-point vehicle passing data corresponding to the first video point, the first candidate vehicle captured within the first target time period is determined, including: for each first passing vehicle, using the several initial captured images and the target captured images corresponding to the first passing vehicle to analyze and determine the final candidate image of the first passing vehicle; based on the final candidate image, the first candidate vehicle captured within the first target time period is selected from each first passing vehicle.

[0009] Among them, a number of initial captured pictures and target captured pictures corresponding to the first passing vehicle are used for analysis to determine the final candidate picture of the first passing vehicle, including: extracting features from the initial captured pictures corresponding to the first passing vehicle to obtain features of each initial picture; and extracting features from the target captured picture to obtain target vehicle picture features corresponding to the target vehicle; calculating the similarity between each initial picture feature and the target vehicle picture feature to obtain the similarity of each picture; and selecting the initial captured picture whose picture similarity meets the similarity requirement as the final candidate picture of the first passing vehicle.

[0010] Among them, the similarity requirement is to be the largest among the similarities of each picture corresponding to the first passing vehicle; and / or, based on the final candidate picture, the first candidate vehicle photographed within the first target time period is selected from each first passing vehicle, including: the first passing vehicle whose picture similarity corresponding to the final candidate picture is greater than the similarity threshold is determined as the first candidate vehicle photographed within the first target time period.

[0011] Among them, the preset requirement is that the number of first candidate vehicles is one.

[0012] The method includes: in response to the first candidate vehicle not meeting the preset requirements, obtaining the second candidate vehicle photographed at the second video point within the corresponding second target time period, wherein the second video point is located after the first video point; in response to the second candidate vehicle meeting the preset requirements, obtaining the second elapsed time difference between the entrance of the target vehicle passing through the entrance and exit and the second video point, and taking the ratio of the distance between the entrance and the second video point to the second elapsed time difference as the second average speed; using the second average speed to update the first target time period, and using the updated first target time period to re-execute the steps of obtaining the first candidate vehicle photographed at the first video point within the first target time period and its subsequent steps.

[0013] Among them, after the first candidate vehicle is associated with the target vehicle, it includes: verifying the vehicle passing data of the target vehicle at the trajectory point of the first video point to obtain a verification result; in response to the verification result that the vehicle passing data of the target vehicle at the trajectory point of the first video point meets the trajectory abnormality condition, deleting the vehicle passing data of the target vehicle at the trajectory point of the first video point.

[0014] Among them, the trajectory abnormality condition is that the target vehicle's trajectory data at the first video point at the first video point at least meets the following two: the first median of a number of time differences is greater than the first preset time, each time difference is the difference between the first time when the target vehicle passes the first video point and each second time, each second time is the time when each vehicle in the overlapping vehicle set passes the first video point, the overlapping vehicle set is the intersection of the first vehicle and the second vehicle, the first vehicle is: the vehicle passing the entrance within the first preset time when the target vehicle passes the entrance, the second vehicle is: the vehicle passing the exit within the first preset time when the target vehicle passes the exit; the first ratio is less than the first preset ratio, the first ratio is the number of passes and the number of all video points. The passing number is the number of times the order in which the time sequence vehicle and the target vehicle pass through each video point is inconsistent with the passing order, the passing order is the same order as the target vehicle and the time sequence vehicle pass through the entrance and exit, and the time sequence vehicle is a vehicle that passes through the entrance and exit in the same order as the target vehicle within the second preset time; the first ratio is less than the second preset ratio, the second ratio is the ratio of the number of lane passing times to the number of all video points, the number of lane passing times is the number of times the lane in which the target vehicle is located when passing through the entrance and exit is the same, and the lane in which it is located when passing through each video point is inconsistent with the lane in which it is located when passing through the entrance and exit; there is an abnormal distribution of the image similarity corresponding to the images taken when the target vehicle passes through each video point.

[0015] Among them, after deleting the trajectory point passing data of the target vehicle at the first video point, it includes: taking the ratio between the trajectory point passing data corresponding to the first video point and the point passing data of all first passing vehicles captured at the first video point as a third ratio; in response to the third ratio being less than the third preset ratio, adjusting the first video point; and / or obtaining the second median of the similarity between the trajectory point passing data and the endpoint passing data of each video point; in response to the existence of an abnormality in the second median, adjusting the video point corresponding to the abnormal second median.

[0016] The second aspect of the present application provides a trajectory restoration and display system, including: a conditional screening page for selecting a target vehicle based on keywords; an information display page for displaying the trajectory point passing data of the target vehicle at each entrance and exit and / or each video point, wherein the trajectory point passing data of the entrance and exit are determined based on the endpoint passing data of the target vehicle photographed at the entrance and exit; the trajectory point passing data of the video point is determined by the endpoint passing data of the target vehicle photographed at the entrance and exit and the point passing data of the target vehicle photographed at several video points located between the entrance and exit, to determine the target time period in which the target vehicle passes through each video point, in response to a first candidate vehicle photographed at the first video point within the first target time period meeting a preset requirement, the first candidate vehicle is associated with the target vehicle, and the point passing data corresponding to the associated vehicle is determined as the trajectory point passing data of the target vehicle at the first video point.

[0017] A third aspect of the present application provides an electronic device, comprising a memory and a processor coupled to each other, wherein the processor is used to execute program instructions stored in the memory to implement the vehicle trajectory restoration method in the first aspect.

[0018] A fourth aspect of the present application provides a computer-readable storage medium having program instructions stored thereon, which, when executed by a processor, implements the vehicle trajectory restoration method in the first aspect.

[0019] The above scheme first obtains the endpoint passing data of the target vehicle photographed at the entrance and exit and the point passing data of the target vehicle photographed at several video points, determines the target time period of the target vehicle passing through each video point through the endpoint passing data and the point passing data, obtains the first candidate vehicle photographed at the first video point within the first target time period, and in response to the first candidate vehicle meeting the preset requirements, associates the first candidate vehicle with the target vehicle, and uses the point passing data corresponding to the associated vehicle as the trajectory point passing data of the target vehicle at the first video point, and by using the inferred target time period of the target vehicle passing through each video point, selects the candidate vehicles that meet the preset requirements from the candidate vehicles within the target time period of each video point and associates them with the target vehicle to determine the trajectory point passing data of the target vehicle at each video point, thereby accurately obtaining the trajectory point passing data of the target vehicle on the road.

[0020] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present application and are used together with the specification to illustrate the technical solution of the present application.

[0022] Figure 1 It is a flow chart of an embodiment of a vehicle trajectory restoration method of the present application;

[0023] Figure 2 This is a flow chart of an embodiment of a method for verifying vehicle data at a track point of the present application;

[0024] Figure 3 It is a flow chart of another embodiment of the vehicle trajectory restoration method of the present application;

[0025] Figure 4 It is a schematic diagram of the framework of an embodiment of the trajectory restoration and display system of the present application;

[0026] Figure 5 It is a schematic diagram of the framework of an embodiment of the vehicle trajectory restoration device of the present application;

[0027] Figure 6 It is a schematic diagram of the framework of an embodiment of the electronic device of the present application;

[0028] Figure 7 It is a schematic diagram of a framework of an embodiment of a computer-readable storage medium of the present application. DETAILED DESCRIPTION

[0029] The scheme of the embodiment of the present application is described in detail below in conjunction with the drawings of the specification.

[0030] In the following description, for the purpose of explanation rather than limitation, specific details such as specific system structures, interfaces, and technologies are provided to facilitate a thorough understanding of the present application.

[0031] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the objects associated before and after are in an "or" relationship. In addition, "many" in this article means two or more than two. In addition, the term "at least one" in this article means any combination of at least two of any one or more of a plurality of, for example, including at least one of A, B, and C, can mean including any one or more elements selected from the set consisting of A, B, and C.

[0032] With the increasing popularity of entrance and exit systems, perception data plays an increasingly important role as an important part of building smart cities. At the same time, on urban highways, only the equipment on the entrance and exit gantry can be used to identify the license plate of the target vehicle to determine whether it is speeding, etc. Other high-speed vehicle shooting equipment cannot accurately identify the license plate, and the cost of installing an entrance and exit system on the highway that can clearly capture the license plate of the target vehicle is high. In addition, due to the long distance between the entrances and exits, it is difficult to accurately determine the information of the vehicle between the entrances and exits, resulting in traffic hazards and the phenomenon of some vehicles evading fees. The traditional method of determining the spatiotemporal information of the vehicle is to use the distance and time difference between the entrances and exits to calculate the speed of the target vehicle to locate the spatiotemporal information of the target vehicle, and then use the view information of the target vehicle for verification. However, this method has the problem of inaccurate spatiotemporal information and data mislabeling caused by multiple similar vehicles passing through the entrances and exits within a period of time.

[0033] In the present application, the trajectory of the vehicle passing data at the video points that meet the target time period is restored based on the vehicle speed using an image-data fusion algorithm, so as to accurately obtain the trajectory point passing data of the target vehicle at each video point.

[0034] See also Figure 1 , Figure 1 This is a flow chart of an embodiment of the vehicle trajectory restoration method of the present application. Specifically, it may include the following steps:

[0035] Step S110: Acquire endpoint passing data of the target vehicle photographed at the entrance and exit and point passing data of the passing vehicle photographed at a plurality of video points, wherein the plurality of video points are all set between the entrance and exit.

[0036] This application is mainly used in the field of big data perception data analysis. By using the endpoint passing data captured by the shooting device when the target vehicle passes through the entrance and exit, and the point passing data captured by the shooting device when the target vehicle passes through several video points, the target time period of the target vehicle passing through each video point is obtained, and then the candidate vehicle passing the video point in the target time period is determined by the target time period corresponding to the video point. If the candidate vehicle meets the preset requirements, it means that the candidate vehicle and the target vehicle are the same vehicle. The candidate vehicle and the target vehicle are associated, and the point passing data of the candidate vehicle at the video point is used as the trajectory point passing data of the target vehicle, so that the time and space information of the target vehicle can be accurately obtained. Among them, the entrance and exit are the exit and entrance to a certain section of road, for example, the exit and entrance of a highway section, and the entrance and exit can also be a checkpoint; the video point represents a position on a certain section of road, and the position uses the shooting device to shoot a video of the vehicle passing through the position.

[0037] In some embodiments, the shooting equipment set at the entrance and exit and the shooting equipment set at the video point can be different shooting equipment. The shooting equipment set at the video point is mainly used to observe the road conditions on the highway, and the video points are densely distributed and the equipment is relatively high. Therefore, the specifications of the shooting equipment set at the video point are relatively low, and it is not necessary to take a clear picture of the license plate number, but the point-by-point vehicle data captured by it must at least include the vehicle shooting time, vehicle brand, vehicle color, vehicle type, vehicle shooting picture view and other information, and the vehicle shooting picture angles and shooting directions of different video points may be different. For example, the shooting equipment set at the video point can be a camera, etc.

[0038] The camera equipment installed at the entrances and exits is required to be able to clearly capture the vehicle's license plate number while the vehicle is moving at high speed. Therefore, the requirements for the camera equipment installed at the entrances and exits are relatively high. In addition, the endpoint vehicle passing data captured by the camera equipment installed at the entrances and exits at least includes information such as the license plate number, vehicle shooting time, vehicle brand, vehicle color, vehicle type, and vehicle shooting image view features. For example, high-definition cameras, speed limit probes, lighting, laser vehicle inspection devices and other camera equipment are installed on the gantries of the highway entrances and exits. In addition, communication equipment, ETC (Electronic Toll Collection) equipment, etc. can also be installed on the gantries of the highway entrances and exits.

[0039] Among them, the video points set between two entrances and exits can be evenly distributed or randomly distributed. For example, the distance between adjacent exits and entrances is 10 kilometers, and the video points are evenly distributed. Therefore, a video point can be set every 500 meters between the exit and the entrance, and a total of 19 video points can be set. For another example, the distance between adjacent exits and entrances is 10 kilometers, and the video points are randomly distributed. The first 10 video points are set every 500 meters, and the 1-kilometer section after that is an accident-prone section. Therefore, a video point is set every 300 meters in the 1-kilometer section, and then a video point is set every 600 meters. Therefore, a total of 19 video points can be set between the exit and the entrance, but the 19 video points are unevenly distributed.

[0040] It is understandable that the setting of the video points can be set according to the actual road conditions and is not specifically limited here.

[0041] Step S120: Based on the endpoint vehicle passing data and the vehicle passing data at each point, determine the target time period for the target vehicle to pass through each video point.

[0042] Among them, at least one video point of the plurality of video points is a first video point, and the target time period corresponding to the first video point is a first target time period.

[0043] In some embodiments, the endpoint vehicle passing data includes the start shooting time, the end shooting time, and the first distance between the entrance and exit. The start shooting time represents the time when the target vehicle passes through the entrance of the entrance and exit, and the end shooting time represents the time when the target vehicle passes through the exit of the entrance and exit. The point vehicle passing data includes the second distance between the entrance and the video point. Therefore, in order to determine the target time period for the target vehicle to pass through each video point, the difference between the start shooting time and the end shooting time can be first obtained as the first passing time difference for the target vehicle to pass through the entrance and exit. Then, the ratio between the first distance and the first passing time difference is obtained as the first average speed of the target vehicle. Based on the first average speed and the second distance, the target time period of the video point is obtained.

[0044] Specifically, based on the endpoint vehicle passing data and the vehicle passing data at each point, the first average speed of the target vehicle between the entrances and exits is obtained: v=s / t, where s is the first distance between the entrances and exits, and t is the first time difference. The ratio between the second distance and the first average speed can be obtained as the predicted driving time of the target vehicle, and then the sum of the predicted driving time and the starting shooting time is obtained as the predicted arrival time. The predicted arrival time is the ideal time, that is, the target vehicle is assumed to maintain the first average speed between the entrances and exits. However, the actual situation is usually different from the default situation. The target vehicle usually does not maintain a speed between the entrances and exits. Usually, there will be acceleration or deceleration. Therefore, in order to accurately infer the target time period when the target vehicle passes the video point, the time difference deviation coefficient can be combined for calculation. Based on the predicted arrival time and the time difference deviation coefficient, the target time period is obtained, where the time difference deviation coefficient is related to the discrete degree characterization value of the first average speed of the target vehicle. The difference between the predicted arrival time and the time difference deviation coefficient is taken as the minimum value of the target time period, and the sum of the predicted arrival time and the time difference deviation coefficient is taken as the maximum value of the target time period.

[0045] For example, the predicted travel time is: ,in, is the second distance. Predicted arrival time: ,in, is the starting shooting time. Then the target time period is , where T is the time difference bias coefficient, which is related to the variance of the first average speed. It can be understood that the time difference bias coefficient can be 3, 5, 10, etc., and is not specifically limited here.

[0046] Through the above method, the time when the target vehicle passes the video point can be quickly determined, so that the point passing data of the target vehicle passing the video point can be determined, without searching and comparing the data captured at the video point to determine the point passing data of the target vehicle.

[0047] In other embodiments, the ratio between the speed of the target vehicle when passing through the entrance and the second distance can be used to determine the predicted arrival time. Compared with the previous embodiment, the method in this embodiment is more real-time and can directly use the speed of the target vehicle when passing through the entrance to predict the time when the target vehicle reaches the video point. At the same time, the method in this embodiment assumes that the speed of the target vehicle between the entrance and the video point remains unchanged, but the actual situation is that the target vehicle may also accelerate or decelerate between the entrance and the video point. Therefore, after the predicted arrival time is obtained, the target vehicle may not pass or have passed the video point within the predicted arrival time, so it is necessary to calculate the target time period in combination with the preset search time. The preset search time and the predicted arrival time are used for calculation to determine the target time period. Specifically, after the predicted arrival time is obtained, the sum of the preset search time and the predicted arrival time is used as the maximum value of the target time period; the difference between the preset search time and the predicted arrival time is used as the minimum value of the target time period. Among them, the preset search time is obtained based on historical experience.

[0048] Step S130: Acquire a first candidate vehicle captured at the first video point within the first target time period.

[0049] In some embodiments, the first candidate vehicle captured within the first target time period may be determined based on the point vehicle passing data and the endpoint vehicle passing data corresponding to the first video point.

[0050] Specifically, the point-by-point vehicle data corresponding to the first video point includes a number of initial photographs corresponding to each first passing vehicle, the first passing vehicle is a vehicle passing the first video point photographed within the first target time period, and the endpoint vehicle data includes a target photograph corresponding to the entrance of the target vehicle passing through the entrance and exit. Therefore, the first candidate vehicle photographed within the first target time period can be determined using a number of initial photographs and target photographs, please refer to steps S131 to S132 in combination.

[0051] Step S131: for each first passing vehicle, a number of initial captured images and target captured images corresponding to the first passing vehicle are used for analysis to determine a final candidate image of the first passing vehicle.

[0052] In some embodiments, feature extraction may be performed on several initial photographs corresponding to the first passing vehicle to obtain the features of each initial photograph. For example, feature extraction may be performed on several initial photographs corresponding to the first passing vehicle to obtain initial photograph features such as vehicle color features, vehicle brand features, vehicle type features, whether it is loaded with goods, and vehicle view features corresponding to the first passing vehicle. At the same time, feature extraction may be performed on the target photographs to obtain target vehicle picture features corresponding to the target vehicle. For example, feature extraction may be performed on the target photographs to obtain target vehicle picture features such as vehicle color features, vehicle brand features, vehicle type features, whether it is loaded with goods, and vehicle view features corresponding to the target vehicle. Among them, a trained convolutional neural network may be used to extract features from the above photographs to obtain picture features. In addition, edge detection, autoencoder, and other methods may also be used to extract features from the photographs, which are not specifically limited here.

[0053] After that, the similarity between each initial image feature and the target vehicle image feature is calculated to obtain the similarity of each image. For example, the vehicle color feature of the first passing vehicle is calculated to obtain the first feature; the vehicle brand feature of the first passing vehicle is calculated to obtain the second feature; ...; the vehicle view feature of the first passing vehicle is calculated to obtain the Nth feature, and the above features are weighted fused to obtain the image similarity between each initial image feature and the target vehicle image. Then, the initial captured image whose image similarity meets the similarity requirement is selected as the final candidate image of the first passing vehicle. Among them, the similarity requirement is to be the largest among the similarities of each image corresponding to the first passing vehicle.

[0054] It is understandable that the method for calculating the similarity between each initial image feature and the target vehicle image feature may adopt methods such as cosine similarity, Pearson correlation coefficient, Manhattan distance, etc., and no specific limitation is made here.

[0055] Step S132: Based on the final candidate images, select the first candidate vehicles photographed within the first target time period from the first passing vehicles.

[0056] In some embodiments, the first passing vehicle whose image similarity corresponding to the final candidate image is greater than the similarity threshold can be determined as the first candidate vehicle photographed within the first target time period. If there is only one first passing vehicle after the similarity threshold screening, then the first passing vehicle is the only first candidate vehicle photographed within the first target time period; if there are multiple first passing vehicles after the similarity threshold screening, then these first passing vehicles are the first candidate vehicles photographed within the first target time period; if there are 0 first passing vehicles after the similarity threshold screening, then the point vehicle data captured at the video point is abnormal.

[0057] Step S140: In response to the first candidate vehicle meeting the preset requirements, the first candidate vehicle is associated with the target vehicle, and the point-by-point vehicle data corresponding to the associated vehicle is used as the trajectory point-by-point vehicle data of the target vehicle at the first video point.

[0058] In some embodiments, big data analysis technology can be used to analyze the passing data of a large number of vehicles to establish a behavior model of the target vehicle. Within a preset time period, the passing data of the first candidate vehicle can be compared with the behavior model of the target vehicle. If the preset requirements are met, the vehicle can be determined to be the target vehicle. The preset requirements can be set as the passing data of the first candidate vehicle and the output result of the behavior model of the target vehicle are within the allowable error range.

[0059] In other embodiments, the preset requirement may be set to one first candidate vehicle. If only one first candidate vehicle is obtained within the first target time period, the first candidate vehicle is directly associated with the target vehicle. In the previous step, it is known that the first candidate vehicle and the target vehicle are highly similar, so in this case, the first candidate vehicle can be directly associated with the target vehicle.

[0060] In addition, if there are multiple first candidate vehicles and they do not meet the preset requirements: In response to the first candidate vehicle not meeting the preset requirements, obtain the second candidate vehicle photographed by the second video point in the corresponding second target time period, wherein the second video point is located after the first video point, and the calculation method of the second target time period is the same as the calculation method of the first target time period, which will not be repeated here. Afterwards, determine the second passing vehicle passing the second video point in the second target time period, and select the second candidate vehicle from each second passing vehicle. The specific method can also refer to the selection method of the first candidate vehicle, which will not be repeated here. In response to the second candidate vehicle meeting the preset requirements, obtain the second passing time difference between the entrance and the second video point in the entrance and exit of the target vehicle, and use the ratio between the distance between the entrance and the second video point and the second passing time difference as the second average speed. Then use the second average speed to update the first target time period, and re-execute step S130 and subsequent steps using the updated first target time period.

[0061] For example, 10 video points are set between the entrances and exits. After executing steps S110 to S130 for video point No. 1, if it is obtained that the number of first candidate vehicles photographed at video point No. 1 within the first target time period is 2, then video point No. 1 is not associated first. Steps S110 to S130 are also performed for a video point after video point No. 1, that is, video point No. 2. If the second candidate vehicle captured by video point No. 2 within the second target time period is one, the second candidate vehicle is associated with the target vehicle, and the second time difference between the entrance of the entrance and video point No. 2 is obtained. The ratio of the distance between the entrance and video point No. 2 and the second time difference is used as the second average speed. The first target time period of video point No. 1 is then updated using the second average speed. Several initial captured images and target captured images corresponding to the new first passing vehicle captured within the updated first target time period are analyzed to determine the final candidate image of the new first passing vehicle. The new first passing vehicle corresponding to the final candidate image whose image similarity is greater than the similarity threshold is determined as the new first candidate vehicle captured within the updated first target time period. At this time, if the new first candidate vehicle is one, the new first candidate vehicle is associated with the target vehicle.

[0062] If there are also multiple second candidate vehicles captured by video point No. 2 within the second target time period, then video point No. 2 is not associated first, and the video point after video point No. 2, that is, video point No. 3, is also executed in the same way as step S110 to step S130, and so on. When there is only one candidate vehicle for a video point, the target time period of the previous video point is updated using the average speed obtained at the video point, and the target vehicle passing data of most or all video points between the entrances and exits can be obtained by progressing forward. Through such a progressive and detailed judgment and processing process, the accuracy of the vehicle passing data association is ensured to the greatest extent.

[0063] In some embodiments, after the vehicle passing data is associated, in order to further ensure the accuracy of the restored target vehicle's track point passing data at each video point, the restored track point passing data of each video point can also be checked. The track point passing data of the video point that passes the check will be uploaded to the track restoration display system, otherwise it will be deleted. For details, please refer to Figure 2 , Figure 2 This is a flow chart of an embodiment of a method for verifying vehicle data at a track point of the present application. Specifically, it may include the following steps:

[0064] Step S141: verify the vehicle passing data of the target vehicle at the trajectory point of the first video point to obtain a verification result.

[0065] In some embodiments, the time difference analysis and verification can be performed on the trajectory point passing vehicle data of the first video point. Specifically, first determine the vehicle within the first preset time of the target vehicle passing through the entrance of the entrance and exit as the first vehicle, and then determine the vehicle that also passes through the exit within the first preset time of the target vehicle passing through the exit of the entrance and exit as the second vehicle, and obtain the intersection between the first vehicle and the second vehicle, so as to obtain the overlapping vehicle set. The first preset time here can be comprehensively determined based on factors such as the usual spacing, speed, and data statistical analysis of vehicles in actual traffic scenarios. Generally, a reasonable time length that can cover both normal related vehicles and effectively exclude irrelevant vehicles is taken. Then analyze the distribution of the time difference between the target vehicle passing through each video point and the vehicles in the overlapping vehicle set. If there is a median of the difference between the first time of the target vehicle passing through the first video point and the time of each vehicle in each overlapping vehicle set passing through the first video point (or other reasonable abnormal analysis methods, such as interquartile range methods) greater than the first preset time, then the verification result obtained is that the trajectory point passing vehicle data of the first video point is abnormal. For example, if the first preset time is set to 10 seconds, and the median time difference between the first time when the target vehicle passes a certain video point and the second time when most vehicles in the overlapping vehicle set pass reaches 15 seconds, then it means that the time difference of the passing data of this trajectory point does not conform to the normal rules, and there may be a problem of inaccurate trajectory restoration.

[0066] In other embodiments, the shooting sequence of the target vehicle can be verified. Specifically, if there is a vehicle that passes through the exit and entrance in the same order as the target vehicle within the second preset time (a time difference range threshold determined according to actual traffic conditions and data analysis, such as 5 seconds, 15 seconds, etc.), then the vehicle is a time sequence vehicle (the time sequence vehicle passes faster than the target vehicle at the entrance and exit, or passes slower than the target vehicle), and the number of times the sequence of the time sequence vehicle and the target vehicle passing through each video point is inconsistent with the sequence when passing through the entrance and exit is obtained as the number of passes. If the ratio between the number of passes and the number of all video points is less than the first preset ratio, the verification result obtained is that the trajectory point passing data corresponding to the video point is abnormal. This is based on the principle that the order of vehicle travel in normal traffic flow is relatively stable. If a shooting situation that does not conform to the normal timing occurs, it is likely to mean that there is an error in the trajectory restoration, which requires further verification and correction.

[0067] In other embodiments, the lane information of the target vehicle can be verified, and by comparing the lane information of the target vehicle at the entrance and exit and the video point, the abnormal situation in the trajectory restoration of the target vehicle can also be effectively identified. Specifically, if the lanes of the target vehicle passing through adjacent exits and entrances are consistent, and at the same time, between the adjacent exits and entrances, the ratio between the number of times the lane where the target vehicle is located when passing through each video point and the lane where it is located when passing through the entrance and exit is inconsistent and the number of all video points is less than a second preset ratio (the ratio can be determined by analyzing a large amount of actual traffic data to a reasonable range, such as less than 5%), then the verification result obtained is that the trajectory point passing data corresponding to those video points with inconsistent lanes is abnormal. Under normal circumstances, when the vehicle does not have a lane change violation or other behaviors, its driving lane should maintain relative continuity. If there are many video point lanes that are inconsistent with the entrance and exit lanes and exceed a reasonable ratio range, it is likely that there is a deviation in the trajectory restoration, resulting in inaccurate corresponding lane information of the vehicle. Such a verification method can timely discover such problems and ensure the rationality of trajectory restoration.

[0068] In other embodiments, the image similarity between the vehicle passing data of the trajectory point corresponding to the video point and the vehicle passing data at the entrance and exit can be analyzed to verify whether the vehicle passing data of the trajectory point corresponding to the video point is abnormal. Specifically, the image similarity between the vehicle passing data of the trajectory point and the vehicle passing data at the entrance and exit can be analyzed using algorithms such as the box line method and the isolation forest, which can effectively determine whether the vehicle passing data of the trajectory point is abnormal. Taking the box line method as an example, the image similarity data of the vehicle passing data of the trajectory point at each video point and the shooting data of the entrance and exit are first sorted out, and its distribution is displayed through a box line diagram, and key indicators such as the interquartile range, upper and lower limits are observed. If there are data points that exceed the reasonable upper and lower limits (the upper and lower limits can be determined according to the normal data distribution and the empirical value), the verification result obtained is that the vehicle passing data of the trajectory point corresponding to the abnormal data point is abnormal. Taking the isolation forest algorithm as an example, if some image similarity data are judged to be isolated points (largely different from most other data), the verification result obtained is that the vehicle passing data of the trajectory point at the video point corresponding to the abnormal data point is abnormal. Through this data analysis method, the accuracy of trajectory restoration is verified from the perspective of image similarity to ensure that the trajectory of the target vehicle finally restored conforms to the actual traffic conditions.

[0069] Step S142: In response to the verification result that the vehicle passing data of the trajectory point of the target vehicle at the first video point meets the trajectory abnormality condition, the vehicle passing data of the trajectory point of the target vehicle at the first video point is deleted.

[0070] In some embodiments, the track abnormality condition is that the track point passing data of the target vehicle at the first video point at least satisfies the following two conditions:

[0071] 1. The first median of a number of time differences is greater than the first preset time, each time difference is the difference between the first time when the target vehicle passes the first video point and each second time, each second time is the time when each vehicle in the overlapping vehicle set passes the first video point, the overlapping vehicle set is the intersection of the first vehicle and the second vehicle, the first vehicle is: the vehicle passing the entrance within the first preset time when the target vehicle passes the entrance of the entrance and exit, the second vehicle is: the vehicle passing the exit within the first preset time when the target vehicle passes the exit of the entrance and exit;

[0072] 2. The first ratio is less than the first preset ratio, which is the ratio between the number of passes and the number of all video points. The number of passes is the number of times the sequence of the time sequence vehicle and the target vehicle passing through each video point is inconsistent with the passing sequence. The passing sequence is the same sequence as the target vehicle and the time sequence vehicle passing through the entrance and exit. The time sequence vehicle is a vehicle that passes through the entrance and exit in the same order as the target vehicle within the second preset time.

[0073] 3. The second ratio is less than the second preset ratio, which is the ratio of the number of lane passing times to the number of all video points. The number of lane passing times is the number of times the lane in which the target vehicle is located when passing through each video point is different from the lane in which it is located when passing through the entrance and exit, when the lane in which the target vehicle is located when passing through the entrance and exit is the same;

[0074] 4. There is an abnormal distribution of the image similarities corresponding to the images taken when the target vehicle passes through each video point.

[0075] In some embodiments, in order to avoid affecting the subsequent trajectory restoration of other target vehicles, after completing the verification of the trajectory point passing data of each video point, the shooting quality of the video point can also be analyzed. Specifically, the ratio between the trajectory point passing data corresponding to the first video point and the point passing data of all first passing vehicles captured by the first video point is used as the third ratio, and in response to the third ratio being less than the third preset ratio, the first video point is adjusted. If the proportion of successfully restored data is low, it means that there are few cases where the existing method can accurately associate the vehicle trajectory at this video point, and there may be problems such as missing data and inaccurate algorithms that affect the trajectory restoration effect. It is necessary to further find the cause and consider improvement measures.

[0076] In addition, the second median of the similarity between the trajectory point vehicle data and the endpoint vehicle data of each video point is obtained. In response to the presence of an abnormality in the second median, the video point corresponding to the abnormal second median is adjusted. This is based on the principle that similarity can reflect the accuracy of vehicle data association. As a representative statistic, the abnormality of the median often means that there are problems with the overall trajectory restoration quality, and it is necessary to focus on and optimize the data and restoration methods of the corresponding video points. Among them, the analysis method of the similarity can be a box-line method, an isolation forest algorithm, etc., which is not specifically limited here.

[0077] See also Figure 3 , Figure 3 This is a flow chart of another embodiment of the vehicle trajectory restoration method of the present application. Specifically, it includes the following steps:

[0078] Step S310: Acquire the endpoint vehicle passing data of the target vehicle photographed at the entrance and exit and the point vehicle passing data of the passing vehicle photographed at the first video point.

[0079] Among them, the entrance and exit are located before the first video point.

[0080] This step is the same as the acquisition method in the above step S110, and will not be repeated here.

[0081] Step S320: Determine the first candidate vehicle at the first video point using the point vehicle passing data and the endpoint vehicle passing data.

[0082] In some embodiments, the point vehicle passing data and the endpoint vehicle passing data are used to determine the first target time period during which the target vehicle passes the first video point, and then a first candidate vehicle is selected from the vehicles captured by the first video point within the first target time period.

[0083] Specifically, the endpoint vehicle passing data includes the entrance arrival speed and the target shot picture of the target vehicle arriving at the entrance, and the point vehicle passing data includes the third distance between the entrance and the first video point, and a number of first initial shot pictures corresponding to each first passing vehicle. Therefore, in order to determine the first target time period when the target vehicle passes the first video point, the ratio between the third distance and the entrance arrival speed can be first obtained as the first predicted travel time of the target vehicle. The first target time period is obtained by using the first predicted arrival time and the time difference deviation coefficient, wherein the time difference deviation coefficient is related to the discrete degree characterization value of the entrance arrival speed of the target vehicle, the difference between the first predicted arrival time and the time difference deviation coefficient is taken as the minimum value of the first target time period, and the sum of the first predicted arrival time and the time difference deviation coefficient is taken as the maximum value of the first target time period.

[0084] For each first passing vehicle, a plurality of first initial photographed images and target photographed images corresponding to the first passing vehicle are analyzed to determine a final candidate image of the first passing vehicle. Then, based on the final candidate image, a first candidate vehicle photographed within the first target time period is selected from each first passing vehicle.

[0085] Step S330: In response to the first candidate vehicle meeting the preset requirements, the first candidate vehicle is associated with the target vehicle, and the point-by-point vehicle data corresponding to the associated vehicle is used as the trajectory point-by-point vehicle data of the target vehicle at the first video point.

[0086] This step is the same as the above step S140 and will not be described again here.

[0087] Step S340: Based on the target vehicle's track point passing data at the first video point, the point passing data and endpoint passing data are used for the second video point to determine the first candidate vehicle at the first video point and its subsequent steps.

[0088] In some embodiments, the target vehicle's trajectory point passing data at the first video point is used as the endpoint passing data in step S320, and the point passing data of the second video point is analyzed therefrom, and so on, the trajectory point passing data between each video point between the entrances and exits can be obtained.

[0089] The present application first obtains the endpoint passing data of the target vehicle photographed at the entrance and exit and the point passing data of the target vehicle photographed at several video points, determines the target time period of the target vehicle passing through each video point through the endpoint passing data and the point passing data, obtains the first candidate vehicle photographed at the first video point within the first target time period, associates the first candidate vehicle with the target vehicle in response to the first candidate vehicle meeting the preset requirements, and uses the point passing data corresponding to the associated vehicle as the trajectory point passing data of the target vehicle at the first video point, and by utilizing the inferred target time period of the target vehicle passing through each video point, selects the candidate vehicles that meet the preset requirements from the candidate vehicles within the target time period of each video point and associates them with the target vehicle to determine the trajectory point passing data of the target vehicle at each video point, thereby accurately obtaining the trajectory point passing data of the target vehicle on the road.

[0090] In addition, this application uses a more sophisticated and flexible verification mechanism. For example, in the time difference analysis verification, the normal distance between vehicles, vehicle speed, and a large amount of data statistical analysis in actual traffic scenarios are comprehensively considered to reasonably determine the time range parameters, so that the analysis of the time difference between the target vehicle and other related vehicles passing through the video points is more accurate, and it can more effectively identify whether there are abnormalities in the data of the passing vehicles at the trajectory points. At the same time, combined with multi-dimensional comprehensive judgments such as lane information verification, shooting timing verification, and image similarity verification, the judgment conditions and parameters set in each verification link have been verified and optimized with a large amount of actual traffic data, and they cooperate and complement each other, greatly reducing the probability of misjudgment and missed judgment, and improving the reliability of the overall trajectory restoration.

[0091] This application not only focuses on the proportion of vehicle data passing through the trajectory point of each video point to the total shooting data, but also focuses on using indicators such as the median of image similarity, and combines advanced algorithms such as the box-line method and the isolation forest method for in-depth analysis to comprehensively evaluate the trajectory restoration effect of the video point from multiple angles. Once it is found that a certain video point has a suspected abnormal median of image similarity or a low proportion of restored data, it can more accurately locate the problem that may affect the quality of trajectory restoration, thereby promoting targeted rectification of the video point and further improving the accuracy of subsequent trajectory restoration.

[0092] Through these improved technical means, the shortcomings of the original image-data fusion technology solution in trajectory restoration accuracy and quality control have been overcome, making this solution able to more stably and accurately restore the vehicle's driving trajectory on the highway in actual application.

[0093] In addition, this application can also calculate the instantaneous speed of the vehicle in each section of the road based on the distance between consecutive video points and the shooting time difference (the calculation method is similar to the average speed formula, but it pays more attention to the speed calculation within short distances and small time intervals), and then combine the speed limit standard of the section of the road and the actual road conditions (such as whether there is construction, weather conditions and other factors that reasonably affect the vehicle speed) to determine whether the vehicle has speeding and underspeed violations, providing accurate data support for traffic law enforcement, and also helping to promptly detect abnormal driving vehicles, take traffic control measures in advance, and ensure traffic safety and traffic order on highways.

[0094] Those skilled in the art will appreciate that, in the above method of specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of the steps should be determined by their functions and possible internal logic.

[0095] See also Figure 4 , Figure 4It is a framework diagram of an embodiment of the track restoration display system 400 of the present application. The track restoration display system 400 includes a conditional filtering page 410 and an information display page 12. The conditional filtering page 410 is used to select the target vehicle based on keywords. The information display page 420 is used to display the track point passing data of the target vehicle at each entrance and exit and / or each video point, wherein the track point passing data of the entrance and exit are determined based on the endpoint passing data of the target vehicle photographed at the entrance and exit; the track point passing data of the video point is determined by the endpoint passing data of the target vehicle photographed at the entrance and exit and the point passing data of the target vehicle photographed at several video points between the entrance and exit, and the target time period of the target vehicle passing each video point is determined. In response to the first candidate vehicle photographed at the first video point within the first target time period meeting the preset requirements, the first candidate vehicle is associated with the target vehicle, and the point passing data corresponding to the associated vehicle is determined as the track point passing data of the target vehicle at the first video point.

[0096] After restoring the target vehicle's trajectory, the target vehicle's trajectory data at each video point will be uploaded to the trajectory restoration display system 400. After logging into the trajectory restoration display system 400, the subject can search based on the time, license plate number, driving direction and other filtering conditions on the condition filtering page 410. After selecting the filtering conditions, the subject can click the query button to search, and the corresponding search results will be displayed on the information display page 420. In addition, if the subject is not satisfied with the filtering conditions, the subject can click the reset button on the condition filtering page 410 to filter again.

[0097] Among them, the content displayed on the information display page 420 includes the shooting data of the highway entrance and exit, such as the license plate number, passing time, lane information, etc. of each vehicle recorded by the checkpoint; the shooting data of the highway entrance and exit, such as the accurate time and vehicle model of the vehicle entering and exiting the highway; and the restored video point shooting data, such as the video point shooting picture, the corresponding vehicle characteristics and other information. Therefore, the object can easily find the relevant data within a specific time period or a specific vehicle on the trajectory restoration display system 400 according to their own needs. In the display form of the information display page 420, the restored vehicle trajectory can be clearly and intuitively displayed, such as presenting the vehicle's driving route on the highway in the form of a map, and marking the information of each key point (gantry, video point, etc.) passed by, so that the object can understand the vehicle's driving trajectory at a glance, and provide intuitive data display and query services for traffic management, accident analysis and other tasks.

[0098] Based on the restored vehicle trajectory, it can be analyzed whether the vehicle appears on a specified road section. If so, the highway fee for that section needs to be paid. Furthermore, the vehicle speed can be calculated based on the distance between consecutive video points and the shooting time difference, and combined with the road conditions, it can be judged whether the vehicle is speeding or underspeeding.

[0099] See also Figure 5 , Figure 5 : It is a schematic diagram of a framework of an embodiment of a vehicle trajectory restoration device 500 of the present application. The vehicle trajectory restoration device 500 includes a first acquisition module 510, a calculation module 520, a second acquisition module 530 and an association module 540. The first acquisition module 510 executes to acquire the endpoint passing data of the target vehicle photographed at the entrance and exit and the point passing data of the passing vehicle photographed at a plurality of video points, wherein the plurality of video points are all set between the entrance and exit. The calculation module 520 executes to determine the target time period for the target vehicle to pass through each video point based on the endpoint passing data and the point passing data, wherein at least one video point of the plurality of video points is the first video point, and the target time period corresponding to the first video point is the first target time period. The second acquisition module 530 executes to acquire the first candidate vehicle photographed at the first video point within the first target time period. The association module 540 executes to associate the first candidate vehicle with the target vehicle in response to the first candidate vehicle meeting the preset requirements, and uses the point passing data corresponding to the associated vehicle as the trajectory point passing data of the target vehicle at the first video point.

[0100] In some embodiments, the calculation module 520 executes endpoint vehicle passing data including a start shooting time, an end shooting time and a first distance between an entrance and an exit, the start shooting time representing the time when the target vehicle passes through an entrance in an entrance and an exit, and the end shooting time representing the time when the target vehicle passes through an exit in an entrance and an exit; point vehicle passing data including a second distance between an entrance and a video point; based on the endpoint vehicle passing data and each point vehicle passing data, determining a target time period for the target vehicle to pass through each video point, including: obtaining the difference between the start shooting time and the end shooting time as a first passing time difference for the target vehicle to pass through the entrance and exit; obtaining the ratio between the first distance and the first passing time difference as a first average speed of the target vehicle; and obtaining the target time period for the video point based on the first average speed and the second distance.

[0101] In some embodiments, the calculation module 520 executes a target time period for the video point based on the first average speed and the second distance, including: obtaining the ratio between the second distance and the first average speed as the predicted driving time of the target vehicle; obtaining the sum of the predicted driving time and the starting shooting time as the predicted arrival time; obtaining the target time period based on the predicted arrival time and the time difference deviation coefficient, wherein the time difference deviation coefficient is related to the discrete degree characterization value of the first average speed of the target vehicle, the difference between the predicted arrival time and the time difference deviation coefficient is taken as the minimum value of the target time period, and the sum of the predicted arrival time and the time difference deviation coefficient is taken as the maximum value of the target time period.

[0102] In some embodiments, the second acquisition module 530 executes acquisition of the first candidate vehicle captured at the first video point within the first target time period, including: determining the first candidate vehicle captured within the first target time period based on the point vehicle passing data and endpoint vehicle passing data corresponding to the first video point.

[0103] In some embodiments, the second acquisition module 530 executes the point passing vehicle data corresponding to the first video point, including several initial captured pictures corresponding to each first passing vehicle, the first passing vehicle is a vehicle passing the first video point captured within the first target time period, and the endpoint passing vehicle data includes the target captured picture corresponding to the entrance of the target vehicle passing through the entrance and exit; based on the point passing vehicle data and the endpoint passing vehicle data corresponding to the first video point, the first candidate vehicle captured within the first target time period is determined, including: for each first passing vehicle, using the several initial captured pictures and the target captured pictures corresponding to the first passing vehicle to analyze and determine the final candidate picture of the first passing vehicle; based on the final candidate picture, the first candidate vehicle captured within the first target time period is selected from each first passing vehicle.

[0104] In some embodiments, the second acquisition module 530 performs an analysis using several initial captured pictures and target captured pictures corresponding to the first passing vehicle to determine the final candidate picture of the first passing vehicle, including: performing feature extraction on several initial captured pictures corresponding to the first passing vehicle to obtain features of each initial picture; and, performing feature extraction on the target captured picture to obtain target vehicle picture features corresponding to the target vehicle; performing similarity calculation on each initial picture feature and the target vehicle picture feature to obtain the similarity of each picture; and selecting the initial captured picture whose picture similarity meets the similarity requirements as the final candidate picture of the first passing vehicle.

[0105] In some embodiments, the second acquisition module 530 executes the similarity requirement to be the largest among the similarities of each picture corresponding to the first passing vehicle; and / or, based on the final candidate picture, selects the first candidate vehicle photographed within the first target time period from each first passing vehicle, including: determining the first passing vehicle whose picture similarity corresponding to the final candidate picture is greater than the similarity threshold as the first candidate vehicle photographed within the first target time period.

[0106] In some embodiments, the association module 540 performs a preset requirement that the number of the first candidate vehicles is one.

[0107] In some embodiments, the association module 540 executes a method including: in response to the first candidate vehicle not meeting the preset requirements, obtaining a second candidate vehicle captured at a second video point within a corresponding second target time period, wherein the second video point is located after the first video point; in response to the second candidate vehicle meeting the preset requirements, obtaining a second elapsed time difference between an entrance and the second video point in the entrance and exit of the target vehicle, and taking a ratio between the distance between the entrance and the second video point and the second elapsed time difference as a second average speed; using the second average speed to update the first target time period, and using the updated first target time period to re-execute the steps of obtaining the first candidate vehicle captured at the first video point within the first target time period and subsequent steps.

[0108] In some embodiments, after the association module 540 associates the first candidate vehicle with the target vehicle, the following steps are performed: verifying the vehicle passing data of the target vehicle at the trajectory point at the first video point to obtain a verification result; in response to the verification result that the vehicle passing data of the target vehicle at the trajectory point at the first video point meets the trajectory abnormality condition, deleting the vehicle passing data of the target vehicle at the trajectory point at the first video point.

[0109] In some embodiments, the association module 540 executes the trajectory abnormality condition that the target vehicle's trajectory point passing data at the first video point satisfies at least the following two: the first median of a number of time differences is greater than the first preset time, each time difference is the difference between the first time when the target vehicle passes the first video point and each second time, each second time is the time when each vehicle in the overlapping vehicle set passes the first video point, the overlapping vehicle set is the intersection of the first vehicle and the second vehicle, the first vehicle is: the vehicle passing the entrance within the first preset time when the target vehicle passes the entrance of the entrance and exit, the second vehicle is: the vehicle passing the exit within the first preset time when the target vehicle passes the exit of the entrance and exit; the first ratio is less than the first preset ratio, and the first ratio is the number of passes. The ratio between the number of lane passes and the number of all video points, the number of passes is the number of times the order in which the sequence vehicles and the target vehicle pass through each video point is inconsistent with the passing order, the passing order is the same order as the target vehicle and the sequence vehicles pass through the entrances and exits, and the sequence vehicles are vehicles that pass through the entrances and exits in the same order as the target vehicle within the second preset time; the second ratio is less than the second preset ratio, the second ratio is the ratio of the number of lane passes to the number of all video points, the number of lane passes is the number of times the lane in which the target vehicle is located when passing through the entrances and exits is the same, and the lane in which it is located when passing through each video point is inconsistent with the lane in which it is located when passing through the entrances and exits; there is an abnormal distribution of the image similarity corresponding to the images taken when the target vehicle passes through each video point.

[0110] In some embodiments, after deleting the trajectory point passing data of the target vehicle at the first video point, the association module 540 executes the following steps: taking the ratio between the trajectory point passing data corresponding to the first video point and the point passing data of all first passing vehicles captured at the first video point as a third ratio; in response to the third ratio being less than a third preset ratio, adjusting the first video point; and / or obtaining a second median of the similarity between the trajectory point passing data and the endpoint passing data of each video point; in response to an abnormality in the second median, adjusting the video point corresponding to the abnormal second median.

[0111] See also Figure 6 , Figure 6 : is a schematic diagram of a framework of an embodiment of an electronic device 60 of the present application. The electronic device 60 includes a memory 61 and a processor 62 coupled to each other, and the processor 62 is used to execute program instructions stored in the memory 61 to implement the steps in any of the above vehicle trajectory restoration method embodiments. In a specific implementation scenario, the electronic device 60 may include but is not limited to: a microcomputer, a server, and in addition, the electronic device 60 may also include a mobile device such as a laptop computer and a tablet computer, which is not limited here.

[0112] Specifically, the processor 62 is used to control itself and the memory 61 to implement the steps in any of the above-mentioned vehicle trajectory restoration method embodiments. The processor 62 can also be called a CPU (Central Processing Unit). The processor 62 may be an integrated circuit chip with signal processing capabilities. The processor 62 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. In addition, the processor 62 can be implemented by an integrated circuit chip.

[0113] See also Figure 7 , Figure 7 Schematic diagram of a framework of an embodiment of a computer-readable storage medium 70 of the present application. The computer-readable storage medium 70 stores program instructions 701 that can be executed by a processor, and the program instructions 701 are used to implement the steps in any of the above-mentioned vehicle trajectory restoration method embodiments.

[0114] In some embodiments, the functions or modules included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0115] The above description of various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced to each other, and for the sake of brevity, they will not be repeated herein.

[0116] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device implementation described above is only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation, such as units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0117] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0118] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of each implementation method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.

Claims

1. A vehicle trajectory restoration method, characterized in that: include: Obtain endpoint passing data of the target vehicle photographed at the entrance and exit and point passing data of the passing vehicle photographed at a plurality of video points, wherein the plurality of video points are all set between the entrance and exit; Based on the endpoint vehicle passing data and the vehicle passing data of each of the points, determine the target time period for the target vehicle to pass through each of the video points, wherein at least one of the plurality of video points is a first video point, and the target time period corresponding to the first video point is a first target time period; Obtaining a first candidate vehicle captured by the first video point within the first target time period; In response to the first candidate vehicle meeting the preset requirement, the first candidate vehicle is associated with the target vehicle, and the point-by-point vehicle data corresponding to the associated vehicle is used as the track-point vehicle data of the target vehicle at the first video point; The method further includes: in response to the first candidate vehicle not meeting the preset requirement, obtaining a second candidate vehicle photographed at a second video point within a corresponding second target time period, wherein the second video point is located after the first video point; in response to the second candidate vehicle meeting the preset requirement, obtaining a second elapsed time difference between an entrance of the entrance and exit and the second video point when the target vehicle passes through the entrance and exit, and taking a ratio between the distance between the entrance and the second video point and the second elapsed time difference as a second average speed; using the second average speed to update the first target time period, and using the updated first target time period to re-execute the steps of obtaining the first candidate vehicle photographed at the first video point within the first target time period and subsequent steps.

2. The method according to claim 1, characterized in that The endpoint vehicle passing data includes a start shooting time, an end shooting time, and a first distance between the entrance and exit, wherein the start shooting time represents the time when the target vehicle passes through the entrance of the entrance and exit, and the end shooting time represents the time when the target vehicle passes through the exit of the entrance and exit; The point vehicle passing data includes a second distance between the entrance and the video point; The step of determining the target time period for the target vehicle to pass through each of the video points based on the endpoint vehicle passing data and each of the point vehicle passing data includes: Obtaining the difference between the start shooting time and the end shooting time as a first passing time difference of the target vehicle passing through the entrance and exit; Acquire a ratio between the first distance and the first elapsed time difference as a first average speed of the target vehicle; A target time period of the video point is obtained based on the first average speed and the second distance.

3. The method according to claim 2, characterized in that The step of obtaining a target time period of the video point based on the first average speed and the second distance includes: Obtaining a ratio between the second distance and the first average speed as a predicted travel time of the target vehicle; Obtaining the sum of the predicted travel time and the starting shooting time as the predicted arrival time; Based on the predicted arrival time and the time difference bias coefficient, the target time period is obtained, wherein the time difference bias coefficient is related to the discrete degree characterization value of the first average speed of the target vehicle, the difference between the predicted arrival time and the time difference bias coefficient is taken as the minimum value of the target time period, and the sum of the predicted arrival time and the time difference bias coefficient is taken as the maximum value of the target time period.

4. The method according to claim 1, characterized in that The obtaining of the first candidate vehicle captured by the first video point within the first target time period includes: Based on the point vehicle passing data corresponding to the first video point and the endpoint vehicle passing data, a first candidate vehicle captured within the first target time period is determined.

5. The method according to claim 4, characterized in that The point vehicle passing data corresponding to the first video point includes a plurality of initial photographs corresponding to each first passing vehicle, the first passing vehicle is a vehicle passing the first video point photographed within the first target time period, and the endpoint vehicle passing data includes a target photograph corresponding to the entrance of the target vehicle passing through the entrance and exit; The determining, based on the point vehicle passing data corresponding to the first video point and the endpoint vehicle passing data, a first candidate vehicle photographed within the first target time period includes: For each of the first passing vehicles, analyzing the plurality of initial captured images and the target captured images corresponding to the first passing vehicles to determine a final candidate image of the first passing vehicle; Based on the final candidate image, a first candidate vehicle photographed within the first target time period is selected from each of the first passing vehicles.

6. The method according to claim 5, characterized in that The analyzing the plurality of initial captured images and the target captured image corresponding to the first passing vehicle to determine a final candidate image of the first passing vehicle includes: Performing feature extraction on the several initial photographs corresponding to the first passing vehicle respectively to obtain features of each initial photograph; and performing feature extraction on the target photograph to obtain target vehicle picture features corresponding to the target vehicle; Calculating the similarity between each of the initial image features and the target vehicle image features to obtain the similarity of each image; The initially captured picture whose picture similarity meets the similarity requirement is selected as the final candidate picture of the first passing vehicle.

7. The method according to claim 6, characterized in that The similarity requirement is to be the largest among the similarities of the pictures corresponding to the first passing vehicle; And / or, selecting, based on the final candidate image, from the first passing vehicles, a first candidate vehicle photographed within the first target time period, comprising: The first passing vehicle whose picture similarity corresponding to the final candidate picture is greater than a similarity threshold is determined as the first candidate vehicle photographed within the first target time period.

8. The method according to claim 1, characterized in that The preset requirement is that the number of the first candidate vehicles is one.

9. The method according to claim 1, characterized in that: After associating the first candidate vehicle with the target vehicle, the method further comprises: Verifying the vehicle passing data of the trajectory point of the target vehicle at the first video point to obtain a verification result; In response to the verification result that the vehicle passing data of the trajectory point of the target vehicle at the first video point meets the trajectory abnormality condition, the vehicle passing data of the trajectory point of the target vehicle at the first video point is deleted.

10. The method according to claim 9, characterized in that The track abnormality condition is that the track point passing data of the target vehicle at the first video point at least meets the following two conditions: A first median of a plurality of time differences is greater than a first preset time, each of the time differences is a difference between a first time when the target vehicle passes through the first video point and each second time, each of the second times is a time when each vehicle in the overlapping vehicle set passes through the first video point, the overlapping vehicle set is an intersection of the first vehicle and the second vehicle, the first vehicle is a vehicle that passes through the entrance within a first preset time when the target vehicle passes through the entrance of the entrance and exit, and the second vehicle is a vehicle that passes through the exit within a first preset time when the target vehicle passes through the exit of the entrance and exit; The first ratio is less than a first preset ratio, the first ratio being the ratio between the number of passes and the number of all the video points, the number of passes being the number of times the sequence vehicle and the target vehicle pass through each of the video points and the pass order being inconsistent, the pass order being the same sequence when the target vehicle and the sequence vehicle pass through the entrance and exit, and the sequence vehicle being a vehicle that passes through the entrance and exit in the same pass order as the target vehicle within the second preset time; The second ratio is less than the second preset ratio, and the second ratio is the ratio of the number of lane passing times to the number of all the video points, and the number of lane passing times is the number of times the lane in which the target vehicle is located when passing through each of the video points is inconsistent with the lane in which it is located when passing through the entrance and exit, when the lane in which the target vehicle is located when passing through the entrance and exit is the same; There is an abnormal distribution of picture similarities corresponding to pictures taken when the target vehicle passes through each of the video points.

11. The method according to claim 9, characterized in that After deleting the vehicle passing data of the track point of the target vehicle at the first video point, the method includes: The ratio between the track point vehicle passing data corresponding to the first video point and the point vehicle passing data of all first passing vehicles captured by the first video point is used as a third ratio; In response to the third ratio being less than a third preset ratio, adjusting the first video point; And / or, obtaining a second median of the similarity between the trajectory point vehicle passing data and the endpoint vehicle passing data of each of the video points; in response to an abnormality in the second median, adjusting the video point corresponding to the abnormal second median.

12. A trajectory restoration display system, characterized in that: include: Conditional filtering page, used to select target vehicles based on keywords; An information display page is used to display the trajectory point passing data of the target vehicle at each entrance and exit and / or each video point, wherein the trajectory point passing data of the entrance and exit is determined based on the endpoint passing data of the target vehicle photographed at the entrance and exit; the trajectory point passing data of the video point is determined by the endpoint passing data of the target vehicle photographed at the entrance and exit and the point passing data of the target vehicle photographed at several video points located between the entrances and exits, to determine the target time period when the target vehicle passes through each of the video points, and in response to a first candidate vehicle photographed at the first video point within the first target time period meeting a preset requirement, the first candidate vehicle is associated with the target vehicle, and the point passing data corresponding to the associated vehicle is used as the trajectory of the target vehicle at the first video point. In response to the first candidate vehicle not meeting the preset requirement, obtaining a second candidate vehicle photographed at a second video point within a corresponding second target time period, wherein the second video point is located after the first video point; in response to the second candidate vehicle meeting the preset requirement, obtaining a second elapsed time difference between an entrance of the entrance and the second video point when the target vehicle passes through the entrance and exit, and taking a ratio between the distance between the entrance and the second video point and the second elapsed time difference as a second average speed; using the second average speed to update the first target time period, and using the updated first target time period to re-execute the target time period for the target vehicle to pass through each of the video points and subsequent steps.

13. An electronic device, characterized in that: It comprises a memory and a processor coupled to each other, wherein the processor is used to execute program instructions stored in the memory to implement the vehicle trajectory restoration method according to any one of claims 1 to 11.

14. A computer-readable storage medium having program instructions stored thereon, characterized in that: When the program instructions are executed by the processor, the vehicle trajectory restoration method described in any one of claims 1 to 11 is implemented.

Images