Traffic abnormal event influence degree evaluation method based on unmanned aerial vehicle monitoring

Abstract

The invention discloses a traffic abnormal event influence degree evaluation method based on unmanned aerial vehicle monitoring, and belongs to the technical field of traffic abnormal event monitoring. In order to accurately evaluate the traffic abnormal event, the method comprises the following steps: acquiring a traffic video by using an unmanned aerial vehicle, and then acquiring a traffic image from the traffic video by adopting a mode of extracting frames from the video; performing image processing to obtain an updated traffic image; training the YOLOv10 deep learning model to obtain a traffic intelligent identification model for identifying traffic abnormal events; carrying out traffic abnormal event identification, and establishing a conversion relation between an image size and an actual size for the image in which the traffic abnormal event is identified; traffic abnormal event influence degree factors based on unmanned aerial vehicle monitoring are set and include an abnormal event direct influence range, a traffic congestion length caused by an abnormal event, a traffic flow, an average vehicle speed and delay time, and a traffic abnormal event influence degree evaluation method based on unmanned aerial vehicle monitoring is constructed. The method is accurate in evaluation.

Description

Technical Field

[0001] The present invention belongs to the technical field of abnormal traffic event monitoring, and in particular relates to a method for evaluating the impact degree of abnormal traffic events based on unmanned aerial vehicle monitoring. Background Art

[0002] Traffic abnormal events broadly cover any situation that may interrupt normal traffic flow and safety. Traffic abnormal events have far-reaching impacts on socio-economic and public safety. First, these events usually lead to traffic congestion, prolong commuting time, increase fuel consumption and air pollution. At the economic level, traffic congestion reduces labor efficiency, increases transportation costs, and affects the timely supply of goods and services, which may inhibit economic growth. Second, traffic abnormal events increase the risk of traffic accidents, threaten the lives of drivers and pedestrians, and increase the burden on medical and emergency services. In addition, long-term traffic congestion may also affect mental health, leading to increased stress and anxiety for drivers and passengers. In terms of social services, traffic jams may hinder the rapid response of emergency service vehicles (such as ambulances, fire trucks and police cars), affecting their rescue efficiency.

[0003] The lack of a detailed analysis of the impact of traffic anomalies can lead to multiple problems. First, decision makers and urban planners may not be able to accurately identify and prioritize the most serious traffic problems, resulting in improper resource allocation. For example, if it is not clear which types of traffic anomalies have the greatest impact on traffic flow, there may be an over-investment in problems that have a smaller impact on traffic flow, while ignoring more urgent problems that need more attention. Second, the lack of a clear analysis of the impact may result in ineffective emergency response measures and the inability to quickly and specifically address major traffic anomalies, thereby amplifying the negative impact of the incident. In addition, the public and policymakers' perception of traffic safety and efficiency may be limited, affecting public support and participation in traffic planning and policies. Ultimately, this may lead to an erosion of public trust and questioning of traffic management capabilities, further affecting policy implementation and effectiveness.

[0004] In the invention patent with application number 202410511325.X and invention name "A method and system for detecting abnormal traffic events based on drone visual perception", traffic videos are collected by drones, data sets are constructed and two-level models, coarse-grained and fine-grained models, are developed to effectively detect abnormal traffic events.

[0005] However, the above content mainly focuses on collecting traffic video data through drones and using the constructed two-level anomaly detection model to identify abnormal traffic events, but does not involve technical measures for image quality enhancement and specific evaluation methods for the impact of abnormal traffic events, which may affect the accuracy and practicality of the final detection. Summary of the invention

[0006] The problem to be solved by the present invention is to accurately evaluate abnormal traffic events, and provide a method for evaluating the impact of abnormal traffic events based on drone monitoring.

[0007] To achieve the above object, the present invention is implemented through the following technical solutions: A method for evaluating the impact of abnormal traffic events based on drone monitoring includes the following steps: S1. Using a drone to collect traffic videos, and then extracting frames from the traffic videos to obtain traffic images; S2. performing image processing on the traffic image obtained from the traffic video to obtain an updated traffic image; S3. Using the updated traffic image obtained in step S2, manually select images of abnormal traffic events including different types of vehicles, train the YOLOv10 deep learning model, and obtain an intelligent traffic recognition model for identifying abnormal traffic events; S4. The updated traffic image obtained in step S2 is input into the intelligent traffic recognition model for identifying abnormal traffic events in step S3, and the image of the abnormal traffic event is recognized to establish a conversion relationship between the image size and the actual size; S5. Based on the images and videos of the abnormal traffic events identified in step S3 and the conversion relationship between the image size and the actual size obtained in step S4, the impact factors of the abnormal traffic events based on drone monitoring are set, including the direct impact range of the abnormal events, the length of the traffic congestion caused by the abnormal events, the traffic flow, the average vehicle speed, and the delay time, and a method for evaluating the impact of abnormal traffic events based on drone monitoring is constructed.

[0008] Furthermore, the specific implementation method of step S2 includes the following steps: S2.1. For the traffic image obtained from the traffic video in step S1, collect the image at pixel coordinates ( a , b ) is represented by the pixel value at the position A ( a , b ),in, a and b is the pixel coordinate in the horizontal direction and the pixel coordinate in the vertical direction, and the pixel gradient is calculated Horizontal difference in the image The vertical difference , the expression is: ; ; in, For the image at coordinates (a +1, b ) pixel value at the position, For the image at coordinates ( a , b +1) pixel value at position; Then calculate the pixel gradient , the calculation formula is: ; Pixel gradient is used to determine the edge strength at image pixels; S2.2. Setting the adjustment factor c ( a , b ), used to determine different locations in the image ( a , b ) pixel update degree, adjustment coefficient c ( a , b ) is calculated as: ; in, is a constant that controls the sensitivity, To control the constant of the gradient component, both constants are determined empirically; S2.3. Update the pixels of the image and obtain the updated image at coordinates ( a , b ) pixel value at position , the calculation formula is: ; in, is the pixel coordinate ( a , b ) The four pixel coordinates that are adjacent in the top, bottom, left, and right directions, Refers to the pixel values ​​of four adjacent pixel coordinates in the four directions of up, down, left, and right.

[0009] Furthermore, the specific implementation method of step S3 includes the following steps: S3.1. Manually select images of traffic anomaly events including different types of vehicles, including vehicle collision, vehicle breakdown, road collapse, mudslide, building collapse, tree breakage, fog, road icing, sudden congestion, long-term congestion, illegal parking, vehicle fire, hazardous material leakage, and road construction; different types of vehicles include sedans, SUVs, vans, vans, trucks, buses, long-distance coaches, engineering vehicles, two-wheelers, cement mixers, tank trucks, and tractors; S3.2. Use a labeling tool to label the images manually selected in step S3.1, and label the location and category of the vehicle for each image, and the area and type of the abnormal traffic event; S3.3. Build a data set, convert the image annotation file obtained in step S3.2 into txt format, and divide the data set into training set, validation set, and test set according to the ratio of 70%, 15%, and 15% for the images and the corresponding image annotation files; S3.4. Use the training set obtained in step S3.3 to train the YOLOv10 model. First, configure an environment suitable for running the YOLOv10 model and set the learning rate, batch size, training cycle, and hyperparameters. Use the loss during training and the performance on the validation set to monitor the training progress and adjust the training parameters. The precision and recall rate parameters are used to evaluate the effects of vehicle identification and abnormal event detection. When the precision and recall rate reach the set value, the training is stopped. At this time, the corresponding model is a traffic intelligent recognition model for identifying abnormal traffic events.

[0010] Furthermore, the specific implementation method of step S4 includes the following steps: S4.1. The updated traffic image obtained in step S2 is input into the traffic intelligent recognition model for identifying abnormal traffic events in step S3 to identify abnormal traffic events and identify the image of abnormal traffic events; S4.2. Obtaining the UAV’s flight altitude data from the UAV’s onboard positioning system h w , get the camera focal length from the camera device j , the width of the sensor k sen ,high h sen , and the width of the image of the traffic anomaly event k img ,high h img ; Calculate the actual length of the image unit pixel corresponding to the traffic abnormality event, the calculation formula is: ; ; in, and are the ground distance in the horizontal direction and the ground distance in the vertical direction of the image unit pixel of the traffic abnormality event respectively; S4.3. Based on the fact that the ground distance of unit pixels in the image collected by drone is equal in horizontal and vertical directions, that is, , then establish the pixel coordinates ( a , b) and actual geographic coordinates The conversion relationship between them is expressed as: ; ; in, are the geographic coordinates of the image reference point, is the pixel coordinate of the image reference point in the image.

[0011] Furthermore, the specific implementation method of step S5 includes the following steps: S5.1. Calculate the direct impact of abnormal events x 1: Based on the images of identified traffic abnormal events, the abnormal event type and the coordinates of the center point of the abnormal event are extracted ( y a , y b ), the length of the abnormal event C y and the width of the event k y , combined with the ground distance of the unit pixel of the image of the traffic abnormal event in the horizontal direction, calculate the direct impact range of the abnormal event x 1. The calculation expression is: ; S5.2. Calculation of traffic jam length x 2: Based on the images of the identified traffic abnormalities, the vehicle type, vehicle center point coordinates ( Q a , Q b ), length of the vehicle C Q and the width of the vehicle k Q , with the coordinates of the center point of the abnormal event ( y a , y b ) is the starting point, and the identified vehicles are sorted in the opposite direction of the driving direction, and are sorted in order of 1, 2, 3, ..., P, where P is the total number of identified vehicles; After sorting i The center point coordinates of each vehicle are recorded as ( Q ai , Q bi ), i=1,2,…,P, calculate the distance between two adjacent vehicles in order R i , the calculation formula is: ; Setting congestion thresholds Rt ,Compare R i and Rt relationship, when When , it indicates that there is no congestion in the area; when When , it indicates that there is congestion in the area; Traffic abnormalities are most likely to occur near the area where congestion occurs. of i The value of is recorded as it , the corresponding center point coordinates of the vehicle are ( Q ait , Q bit ); Combined with the ground distance of the unit pixel of the image of the traffic abnormality event in the horizontal direction , calculate the center point coordinates of the first vehicle ( Q a1 , Q b1 ) and it The coordinates of the center point of the vehicle ( Q ait , Q bit ) is the traffic congestion length x 2. The calculation formula is: ; S5.3. Calculating traffic flow x 3: Based on the extracted video data, the first section DM1 is manually selected, and the traffic intelligent recognition model for identifying abnormal traffic events is used to identify and count the total number of vehicles that travel to the first section DM1 within 1 minute. U ,based on U Calculate traffic flow x 3. The expression is: ; S5.4. Calculation of average vehicle speed x 4: Based on the extracted video data, the first section DM1 and the second section DM2 are manually selected, and the pixel distance between DM1 and DM2 is extracted as V , using the timestamp in the video data, calculate the time required for the vehicle to travel from section DM1 to section DM2 ta , combined with the ground distance of the unit pixel of the image of the traffic abnormality event in the horizontal direction , calculate the average vehicle speed x 4. The expression is: ; S5.5. Calculation of delay time x 5: Based on the extracted video data, the first section DM1 and the second section DM2 are manually selected, and the pixel distance between DM1 and DM2 is extracted as V , using the timestamp in the video data, calculate the time required for the vehicle to travel from section DM1 to section DM2 ta , combined with the ground distance of the unit pixel of the image of the traffic abnormality event in the horizontal direction The design speed of the road section is v , calculate the delay time x 5. The expression is: ; S5.6. Construct an evaluation method for the impact of abnormal traffic events based on drone monitoring: For different traffic abnormalities, the direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of the corresponding abnormal events are evaluated, and the corresponding comprehensive impact degree is judged by experts; The comprehensive impact degree is divided into three categories, namely mild 1, moderate 2, and severe 3. The hyperplane equation is constructed, and the expression is: ; in, , , , , are the fitting coefficients of traffic congestion length, traffic flow, average vehicle speed, and delay time, respectively. is the fitting constant; The direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of abnormal events corresponding to the light comprehensive impact degree are recorded as x 1-1 , x 2-1 , x 3-1 , x 4-1 , x 5-1 ; The direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of the abnormal event corresponding to the medium comprehensive impact degree are recorded as x 1-2 , x 2-2 , x3-2 , x 4-2 , x 5-2 ; The direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of the abnormal event corresponding to the serious comprehensive impact degree are recorded in turn as x 1-3 , x 2-3 , x 3-3 , x 4-3 , x 5-3 ; For mild impact, set the constraints as follows: ; For moderate impact, set the constraints as: ; For severe impacts, set the constraints as follows: ; Set the objective function to ; Combining the above equations, we can get , , , , , The value of , we get the hyperplane equation; The newly extracted data on the direct impact range of abnormal events, traffic congestion length, traffic flow, average vehicle speed, and delay time are input into the hyperplane equation for calculation and compared with the constraints to complete the impact evaluation of abnormal traffic events based on drone monitoring.

[0012] Beneficial effects of the present invention: The present invention discloses a method for evaluating the impact of abnormal traffic events based on drone monitoring, and proposes an image pixel update method for enhancing the quality of drone-collected images. By independently calculating the changes around each pixel, it can adapt to local features and make more detailed adjustments to the details in the image. The method dynamically adjusts the weights and final pixel values ​​according to the actual pixel changes, making the image enhancement more natural, reducing the risk of over-processing, and improving the image quality by dynamically adjusting the pixel values ​​in the image, making the image enhancement more natural.

[0013] The present invention discloses a method for evaluating the impact of abnormal traffic events based on drone monitoring, proposes a method for evaluating the impact of abnormal traffic events, establishes an impact evaluation system, and realizes the classification of the impact of abnormal traffic events. It is of great significance for optimizing traffic management, improving emergency response efficiency, improving traffic planning, enhancing public safety awareness, and supporting data-driven decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a flow chart of a method for evaluating the impact of abnormal traffic events based on drone monitoring according to the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the specific embodiments described are only part of the embodiments of the present invention, rather than all of the specific embodiments. The components of the specific embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations, and the present invention can also have other embodiments.

[0016] Therefore, the following detailed description of the specific embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents the selected specific embodiments of the present invention. Based on the specific embodiments of the present invention, all other specific embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0017] In order to further understand the content, features and effects of the present invention, the following specific implementation methods are given as examples, and the attached Figure 1 The detailed instructions are as follows:

[0018] Embodiment 1: A method for evaluating the impact of abnormal traffic events based on drone monitoring includes the following steps: S1. Using a drone to collect traffic videos, and then extracting frames from the traffic videos to obtain traffic images; S2. performing image processing on the traffic image obtained from the traffic video to obtain an updated traffic image; Furthermore, the specific implementation method of step S2 includes the following steps: S2.1. For the traffic image obtained from the traffic video in step S1, collect the image at pixel coordinates ( a , b ) is represented by the pixel value at the positionA ( a , b ),in, a and b is the pixel coordinate in the horizontal direction and the pixel coordinate in the vertical direction, and the pixel gradient is calculated Horizontal difference in the image The vertical difference , the expression is: ; ; in, For the image at coordinates ( a +1, b ) pixel value at the position, For the image at coordinates ( a , b +1) pixel value at position; Then calculate the pixel gradient , the calculation formula is: ; Pixel gradient is used to determine the edge strength at image pixels; S2.2. Setting the adjustment factor c ( a , b ), used to determine different locations in the image ( a , b ) pixel update degree, adjustment coefficient c ( a , b ) is calculated as: ; in, is a constant that controls the sensitivity, To control the constant of the gradient component, both constants are determined empirically; S2.3. Update the pixels of the image and obtain the updated image at coordinates ( a , b ) pixel value at position , the calculation formula is: ; in, is the pixel coordinate ( a , b ) The four pixel coordinates that are adjacent in the top, bottom, left, and right directions, Refers to the pixel values ​​of four adjacent pixel coordinates in the four directions of up, down, left, and right.

[0019] S3. Using the updated traffic image obtained in step S2, manually select images of abnormal traffic events including different types of vehicles, train the YOLOv10 deep learning model, and obtain an intelligent traffic recognition model for identifying abnormal traffic events; Furthermore, the specific implementation method of step S3 includes the following steps: S3.1. Manually select images of traffic anomaly events including different types of vehicles, including vehicle collision, vehicle breakdown, road collapse, mudslide, building collapse, tree breakage, fog, road icing, sudden congestion, long-term congestion, illegal parking, vehicle fire, hazardous material leakage, and road construction; different types of vehicles include sedans, SUVs, vans, vans, trucks, buses, long-distance coaches, engineering vehicles, two-wheelers, cement mixers, tank trucks, and tractors; S3.2. Use a labeling tool to label the images manually selected in step S3.1, and label the location and category of the vehicle for each image, and the area and type of the abnormal traffic event; S3.3. Build a data set, convert the image annotation file obtained in step S3.2 into txt format, and divide the data set into training set, validation set, and test set according to the ratio of 70%, 15%, and 15% for the images and the corresponding image annotation files; S3.4. Use the training set obtained in step S3.3 to train the YOLOv10 model. First, configure an environment suitable for running the YOLOv10 model and set the learning rate, batch size, training cycle, and hyperparameters. Use the loss during training and the performance on the validation set to monitor the training progress and adjust the training parameters. The precision and recall rate parameters are used to evaluate the effects of vehicle identification and abnormal event detection. When the precision and recall rate reach the set value, the training is stopped. At this time, the corresponding model is a traffic intelligent recognition model for identifying abnormal traffic events.

[0020] S4. The updated traffic image obtained in step S2 is input into the intelligent traffic recognition model for identifying abnormal traffic events in step S3, and the image of the abnormal traffic event is recognized to establish a conversion relationship between the image size and the actual size; Furthermore, the specific implementation method of step S4 includes the following steps: S4.1. The updated traffic image obtained in step S2 is input into the traffic intelligent recognition model for identifying abnormal traffic events in step S3 to identify abnormal traffic events and identify the image of abnormal traffic events; S4.2. Obtaining the UAV’s flight altitude data from the UAV’s onboard positioning system h w , get the camera focal length from the camera devicej , the width of the sensor k sen ,high h sen , and the width of the image of the traffic anomaly event k img ,high h img ; Calculate the actual length of the image unit pixel corresponding to the traffic abnormality event, the calculation formula is: ; ; in, and are the ground distance in the horizontal direction and the ground distance in the vertical direction of the image unit pixel of the traffic abnormality event respectively; S4.3. Based on the fact that the ground distance of unit pixels in the image collected by drone is equal in horizontal and vertical directions, that is, , then establish the pixel coordinates ( a , b ) and actual geographic coordinates The conversion relationship between them is expressed as: ; ; in, are the geographic coordinates of the image reference point, is the pixel coordinate of the image reference point in the image.

[0021] S5. Based on the images and videos of the abnormal traffic events identified in step S3 and the conversion relationship between the image size and the actual size obtained in step S4, the impact factors of the abnormal traffic events based on drone monitoring are set, including the direct impact range of the abnormal events, the length of the traffic congestion caused by the abnormal events, the traffic flow, the average vehicle speed, and the delay time, and a method for evaluating the impact of abnormal traffic events based on drone monitoring is constructed.

[0022] Furthermore, the specific implementation method of step S5 includes the following steps: S5.1. Calculate the direct impact of abnormal events x 1: Based on the images of identified traffic abnormal events, the abnormal event type and the coordinates of the center point of the abnormal event are extracted ( y a , y b ), the length of the abnormal event C y and the width of the event k y, combined with the ground distance of the unit pixel of the image of the traffic abnormal event in the horizontal direction, calculate the direct impact range of the abnormal event x 1. The calculation expression is: ; S5.2. Calculation of traffic jam length x 2: Based on the images of the identified traffic abnormalities, the vehicle type, vehicle center point coordinates ( Q a , Q b ), length of the vehicle C Q and the width of the vehicle k Q , with the coordinates of the center point of the abnormal event ( y a , y b ) is the starting point, and the identified vehicles are sorted in the opposite direction of the driving direction, and are sorted in order of 1, 2, 3, ..., P, where P is the total number of identified vehicles; After sorting i The center point coordinates of each vehicle are recorded as ( Q ai , Q bi ), i=1,2,…,P, calculate the distance between two adjacent vehicles in order R i , the calculation formula is: ; Setting congestion thresholds Rt ,Compare R i and Rt relationship, when When , it indicates that there is no congestion in the area; when When , it indicates that there is congestion in the area; Traffic abnormalities are most likely to occur near the area where congestion occurs. of i The value of is recorded as it , the corresponding center point coordinates of the vehicle are ( Q ait , Q bit ); Combined with the ground distance of the unit pixel of the image of the traffic abnormality event in the horizontal direction , calculate the center point coordinates of the first vehicle ( Qa1 , Q b1 ) and it The coordinates of the center point of the vehicle ( Q ait , Q bit ) is the traffic congestion length x 2. The calculation formula is: ; S5.3. Calculating traffic flow x 3: Based on the extracted video data, the first section DM1 is manually selected, and the traffic intelligent recognition model for identifying abnormal traffic events is used to identify and count the total number of vehicles that travel to the first section DM1 within 1 minute. U ,based on U Calculate traffic flow x 3. The expression is: ; S5.4. Calculation of average vehicle speed x 4: Based on the extracted video data, the first section DM1 and the second section DM2 are manually selected, and the pixel distance between DM1 and DM2 is extracted as V , using the timestamp in the video data, calculate the time required for the vehicle to travel from section DM1 to section DM2 ta , combined with the ground distance of the unit pixel of the image of the traffic abnormality event in the horizontal direction , calculate the average speed x 4. The expression is: ; S5.5. Calculation of delay time x 5: Based on the extracted video data, the first section DM1 and the second section DM2 are manually selected, and the pixel distance between DM1 and DM2 is extracted as V , using the timestamp in the video data, calculate the time required for the vehicle to travel from section DM1 to section DM2 ta , combined with the ground distance of the unit pixel of the image of the traffic abnormality event in the horizontal direction The design speed of the road section is v , calculate the delay time x 5. The expression is: ; S5.6. Construct an evaluation method for the impact of abnormal traffic events based on drone monitoring: For different traffic abnormalities, the direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of the corresponding abnormal events are evaluated, and the corresponding comprehensive impact degree is judged by experts; The comprehensive impact degree is divided into three categories, namely mild 1, moderate 2, and severe 3. The hyperplane equation is constructed, and the expression is: ; in, , , , , are the fitting coefficients of traffic congestion length, traffic flow, average vehicle speed, and delay time, respectively. is the fitting constant; The direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of abnormal events corresponding to the light comprehensive impact degree are recorded as x 1-1 , x 2-1 , x 3-1 , x 4-1 , x 5-1 ; The direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of the abnormal event corresponding to the medium comprehensive impact degree are recorded as x 1-2 , x 2-2 , x 3-2 , x 4-2 , x 5-2 ; The direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of the abnormal event corresponding to the serious comprehensive impact degree are recorded in sequence as x 1-3 , x 2-3 , x 3-3 , x 4-3 , x 5-3 ; For mild impact, set the constraints as follows: ; For moderate impact, set the constraints as: ; For severe impacts, set the constraints as follows: ; Set the objective function to ; Combining the above equations, we can get , , , , , The value of , we get the hyperplane equation; The newly extracted data on the direct impact range of abnormal events, traffic congestion length, traffic flow, average vehicle speed, and delay time are input into the hyperplane equation for calculation and compared with the constraints to complete the impact evaluation of abnormal traffic events based on drone monitoring.

[0023] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0024] Although the present application has been described above with reference to specific embodiments, various modifications may be made thereto and parts thereof may be replaced with equivalents without departing from the scope of the present application. In particular, as long as there is no structural conflict, the various features in the specific embodiments disclosed in the present application may be used in combination with each other in any manner, and the fact that these combinations are not exhaustively described in this specification is only for the sake of omitting space and saving resources. Therefore, the present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for evaluating the impact of abnormal traffic events based on drone monitoring, characterized in that: The steps include: S1. Using a drone to collect traffic videos, and then extracting frames from the traffic videos to obtain traffic images; S2. performing image processing on the traffic image obtained from the traffic video to obtain an updated traffic image; S3. Using the updated traffic image obtained in step S2, manually select images of abnormal traffic events including different types of vehicles, train the YOLOv10 deep learning model, and obtain an intelligent traffic recognition model for identifying abnormal traffic events; S4. The updated traffic image obtained in step S2 is input into the intelligent traffic recognition model for identifying abnormal traffic events in step S3, and the image of the abnormal traffic event is recognized to establish a conversion relationship between the image size and the actual size; S5. Based on the images and videos of the abnormal traffic events identified in step S3 and the conversion relationship between the image size and the actual size obtained in step S4, the impact factors of the abnormal traffic events based on drone monitoring are set, including the direct impact range of the abnormal events, the length of the traffic congestion caused by the abnormal events, the traffic flow, the average vehicle speed, and the delay time, and a method for evaluating the impact of abnormal traffic events based on drone monitoring is constructed.

2. The method for evaluating the impact of abnormal traffic events based on drone monitoring according to claim 1 is characterized in that: The specific implementation method of step S2 includes the following steps: S2.

1. For the traffic image obtained from the traffic video in step S1, collect the image at pixel coordinates ( a , b ) is represented by the pixel value at the position A ( a , b ),in, a and b is the pixel coordinate in the horizontal direction and the pixel coordinate in the vertical direction, and the pixel gradient is calculated Horizontal difference in the image The vertical difference , the expression is: ； ； in, For the image at coordinates ( a +1, b ) pixel value at the position, For the image at coordinates ( a , b +1) pixel value at position; Then calculate the pixel gradient , the calculation formula is: ； Pixel gradient is used to determine the edge strength at image pixels; S2.

2. Setting the adjustment factor c ( a , b ), used to determine different locations in the image ( a , b ) pixel update degree, adjustment coefficient c ( a , b ) is calculated as: ； in, is a constant that controls the sensitivity, To control the constant of the gradient component, both constants are determined empirically; S2.

3. Update the pixels of the image and obtain the updated image at coordinates ( a , b ) pixel value at position , the calculation formula is: ； in, is the pixel coordinate ( a , b ) The four pixel coordinates that are adjacent in the top, bottom, left, and right directions, Refers to the pixel values ​​of four adjacent pixel coordinates in the four directions of up, down, left, and right.

3. The method for evaluating the impact of abnormal traffic events based on drone monitoring according to claim 2 is characterized in that: The specific implementation method of step S3 includes the following steps: S3.

1. Manually select images of traffic anomaly events including different types of vehicles, including vehicle collision, vehicle breakdown, road collapse, mudslide, building collapse, tree breakage, fog, road icing, sudden congestion, long-term congestion, illegal parking, vehicle fire, hazardous material leakage, and road construction; different types of vehicles include sedans, SUVs, vans, vans, trucks, buses, long-distance coaches, engineering vehicles, two-wheelers, cement mixers, tank trucks, and tractors; S3.

2. Use a labeling tool to label the images manually selected in step S3.1, and label the location and category of the vehicle for each image, and the area and type of the abnormal traffic event; S3.

3. Build a data set, convert the image annotation file obtained in step S3.2 into txt format, and divide the data set into training set, validation set, and test set according to the ratio of 70%, 15%, and 15% for the images and the corresponding image annotation files; S3.

4. Use the training set obtained in step S3.3 to train the YOLOv10 model. First, configure an environment suitable for running the YOLOv10 model and set the learning rate, batch size, training cycle, and hyperparameters. Use the loss during training and the performance on the validation set to monitor the training progress and adjust the training parameters. The precision and recall rate parameters are used to evaluate the effects of vehicle identification and abnormal event detection. When the precision and recall rate reach the set value, the training is stopped. At this time, the corresponding model is a traffic intelligent recognition model for identifying abnormal traffic events.

4. The method for evaluating the impact of abnormal traffic events based on drone monitoring according to claim 3 is characterized in that: The specific implementation method of step S4 includes the following steps: S4.

1. The updated traffic image obtained in step S2 is input into the traffic intelligent recognition model for identifying abnormal traffic events in step S3 to identify abnormal traffic events and identify the image of abnormal traffic events; S4.

2. Obtaining the UAV’s flight altitude data from the UAV’s onboard positioning system h w , get the camera focal length from the camera device j , the width of the sensor k sen ,high h sen , and the width of the image of the traffic anomaly event k img ,high h img ; Calculate the actual length of the image unit pixel corresponding to the traffic abnormality event, the calculation formula is: ； ； in, and are the ground distance in the horizontal direction and the ground distance in the vertical direction of the image unit pixel of the traffic abnormality event respectively; S4.

3. Based on the fact that the ground distance of unit pixels in the image collected by drone is equal in horizontal and vertical directions, that is, , then establish the pixel coordinates ( a , b ) and actual geographic coordinates The conversion relationship between them is expressed as: ； ； in, are the geographic coordinates of the image reference point, is the pixel coordinate of the image reference point in the image.

5. The method for evaluating the impact of abnormal traffic events based on drone monitoring according to claim 4 is characterized in that: The specific implementation method of step S5 includes the following steps: S5.

1. Calculate the direct impact of abnormal events x 1: Based on the images of identified traffic abnormal events, the abnormal event type and the coordinates of the center point of the abnormal event are extracted ( y a , y b ), the length of the abnormal event C y and the width of the event k y , combined with the ground distance of the unit pixel of the image of the traffic abnormal event in the horizontal direction, calculate the direct impact range of the abnormal event x 1. The calculation expression is: ； S5.

2. Calculation of traffic jam length x 2: Based on the images of the identified traffic abnormalities, the vehicle type, vehicle center point coordinates ( Q a , Q b ), length of the vehicle C Q and the width of the vehicle k Q , with the coordinates of the center point of the abnormal event ( y a , y b ) is the starting point, and the identified vehicles are sorted in the opposite direction of the driving direction, and are sorted in order of 1, 2, 3, ..., P, where P is the total number of identified vehicles; After sorting i The center point coordinates of each vehicle are recorded as ( Q ai , Q bi ), i=1,2,…,P, calculate the distance between two adjacent vehicles in order R i , the calculation formula is: ； Setting congestion thresholds R ,Compare R i and R relationship, when When , it indicates that there is no congestion in the area; when When , it indicates that there is congestion in the area; Traffic abnormalities are most likely to occur near the area where congestion occurs. of i The value of is recorded as it , the corresponding center point coordinates of the vehicle are ( Q ait , Q bit ); Combined with the ground distance of the unit pixel of the image of the traffic abnormality event in the horizontal direction , calculate the center point coordinates of the first vehicle ( Q a1 , Q b1 ) and it The coordinates of the center point of the vehicle ( Q ait , Q bit ) is the traffic congestion length x 2. The calculation formula is: ； S5.

3. Calculating traffic flow x 3: Based on the extracted video data, the first section DM1 is manually selected, and the traffic intelligent recognition model for identifying abnormal traffic events is used to identify and count the total number of vehicles that travel to the first section DM1 within 1 minute. U ,based on U Calculate traffic flow x 3. The expression is: ； S5.

4. Calculation of average vehicle speed x 4: Based on the extracted video data, the first section DM1 and the second section DM2 are manually selected, and the pixel distance between DM1 and DM2 is extracted as V , using the timestamp in the video data, calculate the time required for the vehicle to travel from section DM1 to section DM2 ta , combined with the ground distance of the unit pixel of the image of the traffic abnormality event in the horizontal direction , calculate the average vehicle speed x 4. The expression is: ； S5.

5. Calculation of delay time x 5: Based on the extracted video data, the first section DM1 and the second section DM2 are manually selected, and the pixel distance between DM1 and DM2 is extracted as V , using the timestamp in the video data, calculate the time required for the vehicle to travel from section DM1 to section DM2 ta , combined with the ground distance of the unit pixel of the image of the traffic abnormality event in the horizontal direction The design speed of the road section is v , calculate the delay time x 5. The expression is: ； S5.

6. Construct an evaluation method for the impact of abnormal traffic events based on drone monitoring: For different traffic abnormalities, the direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of the corresponding abnormal events are evaluated, and the corresponding comprehensive impact degree is judged by experts; The comprehensive impact degree is divided into three categories, namely mild 1, moderate 2, and severe 3. The hyperplane equation is constructed, and the expression is: ； in, , , , , are the fitting coefficients of traffic congestion length, traffic flow, average vehicle speed, and delay time, respectively. is the fitting constant; The direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of abnormal events corresponding to the light comprehensive impact degree are recorded as x 1-1 , x 2-1 , x 3-1 , x 4-1 , x 5-1 ; The direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of the abnormal event corresponding to the medium comprehensive impact degree are recorded as x 1-2 , x 2-2 , x 3-2 , x 4-2 , x 5-2 ; The direct impact range, traffic jam length, traffic flow, average vehicle speed, and delay time of the abnormal event corresponding to the serious comprehensive impact degree are recorded in sequence as x 1-3 , x 2-3 , x 3-3 , x 4-3 , x 5-3 ; For mild impact, set the constraints as follows: ； For moderate impact, set the constraints as: ； For severe impacts, set the constraints as follows: ； Set the objective function to ; Combining the above equations, we can get , , , , , The value of , we get the hyperplane equation; The newly extracted data on the direct impact range of abnormal events, traffic congestion length, traffic flow, average vehicle speed, and delay time are input into the hyperplane equation for calculation and compared with the constraints to complete the impact evaluation of abnormal traffic events based on drone monitoring.

Images