A highway tunnel video monitoring method and system

Abstract

The application relates to the technical field of tunnel video monitoring and early warning, in particular to a highway tunnel video monitoring method and system, which comprises the following steps: acquiring the temperature, humidity and visibility in a highway tunnel in real time, and monitoring video in the highway tunnel; determining the ring reference degree of the current highway tunnel; extracting and tracking each vehicle in the monitoring video of the highway tunnel through a neural network model; determining the visual field covering interference degree, rear-end collision risk degree, vehicle speed fast-trend coefficient, deceleration necessity degree and deceleration urgency degree of each vehicle; extracting a fault vehicle in the monitoring video of the highway tunnel, determining the deceleration early warning value of each vehicle in the tunnel, and early warning the deceleration of each vehicle in the current tunnel. Therefore, the accuracy of highway video monitoring and early warning is improved.

Description

Technical Field

[0001] This application relates to the field of tunnel video surveillance and early warning technology, specifically to a method and system for video surveillance of highway tunnels. Background Technology

[0002] Highway tunnels play a vital role in the transportation sector. They significantly improve traffic efficiency by shortening vehicle routes and traversing complex terrain. However, the enclosed environment within tunnels increases the risk of accidents. Currently, video surveillance is commonly used to monitor vehicle conditions within tunnels. The tunnel's electromechanical system analyzes the video footage and, based on the analysis, issues warnings to vehicles that need to slow down, thus improving the safety level of vehicle passage within tunnels.

[0003] Traditional video surveillance of highway tunnels typically involves identifying and locating vehicles within the tunnel using video algorithms, then issuing deceleration warnings based on the distance between the current vehicle and the vehicle in front. However, in real-world tunnel scenarios, vehicle conditions and the environment are highly variable. For example, differences in longitudinal speed between vehicles can influence the likelihood of rear-end collisions, while lateral deviations can cause the vehicle in front to obstruct the view of the vehicle behind, increasing driving safety risks. Furthermore, the complex vehicle conditions result in insufficient accuracy in monitoring and analysis results based solely on vehicle distance. Additionally, the tunnel environment varies at different times; in low-temperature and high-humidity conditions, the road surface is prone to icing, further increasing the error rate of traditional deceleration warnings based solely on vehicle distance, thus reducing the accuracy of video surveillance in highway tunnels. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide a method and system for video surveillance of highway tunnels, the specific technical solution of which is as follows:

[0005] In a first aspect, embodiments of this application provide a video surveillance method for highway tunnels, the method comprising the following steps:

[0006] Real-time acquisition of temperature, humidity, visibility, and surveillance video inside highway tunnels;

[0007] Based on the current temperature and humidity ratio inside the highway tunnel, and the difference between the visibility and historical visibility, the environmental parameter optimization of the current highway tunnel is determined.

[0008] The video footage of a highway tunnel was extracted and tracked using a neural network model. The width difference between each vehicle and its neighboring vehicles, as well as the lateral distance between the vehicle centers, were analyzed to determine the visual obstruction interference of each vehicle.

[0009] The rear-end collision risk of each vehicle is determined based on the speed and acceleration differences between each vehicle and its neighboring vehicles; the speed acceleration coefficient of each vehicle is determined based on the length differences between each vehicle and its neighboring vehicles, combined with the ring parameter dominance and the rear-end collision risk.

[0010] By combining the vehicle speed acceleration coefficient and the field of vision obstruction interference, the deceleration necessity of each vehicle is obtained. Combined with the density of vehicles in front of each vehicle in the tunnel, the deceleration urgency of each vehicle is determined.

[0011] Extract the faulty vehicle from the surveillance video of the highway tunnel, analyze the measured distance between each vehicle and the faulty vehicle in the tunnel, as well as the interval between vehicles, and combine the aforementioned deceleration urgency to determine the deceleration warning value for each vehicle in the tunnel, and issue deceleration warnings to each vehicle in the current tunnel.

[0012] In one embodiment, determining the ring parameter dominance includes:

[0013] The average visibility within the tunnel during a preset historical time period is determined, and the difference between the current visibility within the tunnel and the average is calculated. The environmental parameter goodness is positively correlated with the difference and the temperature-humidity ratio.

[0014] In one embodiment, determining the field-of-view occlusion interference degree includes:

[0015] Calculate the ratio of the width of each vehicle to its adjacent vehicle in front, and the field of view occlusion interference is the product of the ratio and the lateral spacing.

[0016] In one embodiment, determining the rear-end collision risk level includes:

[0017] The ratio of the speed of each vehicle to the speed of its nearest preceding vehicle is calculated and denoted as the first ratio. The normalized result of the difference in acceleration between each vehicle and its nearest preceding vehicle is calculated. The rear-end collision risk is the product of the first ratio and the normalized result.

[0018] In one embodiment, determining the vehicle speed acceleration coefficient includes:

[0019] Calculate the average length of each vehicle and its neighboring vehicle in front, and record it as the first average. Determine the longitudinal distance between the center point of each vehicle and the center point of its neighboring vehicle in front. Calculate the ratio of the first average to the longitudinal distance, and record it as the second ratio.

[0020] The reciprocal of the ring parameter optimization is determined, and the vehicle speed acceleration coefficient is the normalized result of the product of the second ratio, the reciprocal, and the rear-end collision risk degree.

[0021] In one embodiment, the deceleration necessity is the square root of the sum of the square of the vehicle speed acceleration coefficient and the square of the field of vision obstruction interference degree.

[0022] In one embodiment, determining the urgency of deceleration includes:

[0023] The ratio of the number of vehicles ahead of each vehicle in the tunnel to the distance to the tunnel exit is calculated and denoted as the third ratio. The urgency of deceleration is the product of the third ratio and the necessity of deceleration.

[0024] In one embodiment, determining the deceleration warning value includes:

[0025] The normalized value of the number of vehicles at intervals and the reciprocal of the measured distance is calculated. The deceleration warning value is positively correlated with both the normalized value and the deceleration urgency.

[0026] In one embodiment, the step of providing deceleration warnings to vehicles currently in the tunnel includes:

[0027] Calculate the normalized value of the difference between the current deceleration warning value of each vehicle in the tunnel and the average deceleration warning value of all vehicles passing through the tunnel in the current preset historical time period. If the normalized value is greater than a preset threshold, the corresponding vehicle is designated as a vehicle requiring a warning; otherwise, the corresponding vehicle is not designated as a vehicle requiring a warning.

[0028] Secondly, embodiments of this application also provide a video surveillance system for highway tunnels, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0029] This application has at least the following beneficial effects:

[0030] This application acquires real-time temperature, humidity, and visibility data within highway tunnels. Based on the current temperature-humidity ratio and the difference between current and historical visibility, it determines the environmental parameter superiority of the current highway tunnel. This determination quantifies the health of the tunnel environment in real time, improving the dynamic adaptability of environmental risk identification and enhancing the predictive ability of the correlation between the environment and accident warnings. A neural network model is used to extract and track vehicles within the monitoring video of the highway tunnel. The width differences between each vehicle and its neighboring vehicles, as well as the lateral distance between vehicle centers, are analyzed to determine the visual obstruction interference degree of each vehicle. This determination improves the perception accuracy of blind spots during driving, provides data support for vehicle spacing control and early warning within tunnels, and optimizes the rationality of lane resource allocation. Based on the speed and acceleration differences between each vehicle and its neighboring vehicles, the rear-end collision risk degree of each vehicle is determined. This rear-end collision risk degree strengthens the forward-looking prediction of collision risks, avoids passive responses to rear-end collisions, and improves the granularity of vehicle risk positioning, which is helpful for targeted vehicle collisions. This method involves issuing early warning information; determining the speed acceleration coefficient of each vehicle based on the length difference between each vehicle and its neighboring vehicles, combined with the ring parameter superiority and the rear-end collision risk; integrating the speed acceleration coefficient with the field of vision obstruction interference to obtain the deceleration necessity of each vehicle; and determining the deceleration urgency of each vehicle by combining the density of vehicles in front of each vehicle in the tunnel. This improves the comprehensiveness of vehicle risk warning decisions in tunnels, avoids misjudgments caused by single-indicator decisions, optimizes the allocation efficiency of early warning resources, and enhances the differentiation capability of tunnel early warning by calculating the deceleration urgency, which helps to improve the priority of emergency response. The method also involves extracting faulty vehicles from the monitoring video of highway tunnels, analyzing the metric distance between each vehicle and the faulty vehicle in the tunnel, as well as the interval vehicles, and combining the deceleration urgency to determine the deceleration warning value for each vehicle in the tunnel. This method provides deceleration warnings to each vehicle in the current tunnel. By realizing the dynamic generation of deceleration warning values, this method integrates multiple influencing factors of tunnel early warning, improves the accuracy of tunnel early warning, optimizes tunnel traffic efficiency, and enhances tunnel traffic safety. Attached Figure Description

[0031] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 A flowchart illustrating the steps of a video surveillance method for highway tunnels, as provided in one embodiment of this application;

[0033] Figure 2This is a schematic diagram of lateral offset displacement;

[0034] Figure 3 Flowchart for determining the deceleration warning value for vehicles inside the tunnel. Detailed Implementation

[0035] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a video monitoring method and system for highway tunnels proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0037] The following description, in conjunction with the accompanying drawings, details the specific scheme of a video surveillance method and system for highway tunnels provided in this application.

[0038] Please see Figure 1 The diagram illustrates a flowchart of a video surveillance method for highway tunnels according to an embodiment of this application. The method includes the following steps:

[0039] S1 can acquire real-time temperature, humidity, visibility, and surveillance video of the highway tunnel.

[0040] The enclosed environment inside highway tunnels, especially under complex traffic conditions, greatly increases the probability of vehicle accidents. Therefore, video footage of vehicle conditions inside the tunnel is often collected, and then the tunnel's electromechanical system analyzes and monitors the vehicles in the video.

[0041] Traditional tunnel electromechanical systems often assess vehicle deceleration warnings based on the distance between vehicles. However, in real-world scenarios, the tunnel environment and vehicle conditions are complex and ever-changing, making the monitoring accuracy based solely on vehicle distance insufficient. Therefore, this embodiment combines the actual tunnel environment and complex vehicle conditions in a real-world scenario for specific analysis, obtaining more accurate deceleration warning assessment results for each vehicle from the tunnel electromechanical system. This reduces vehicle driving risks and improves the accuracy of video monitoring for highway tunnel warnings.

[0042] First, the monitoring video data of vehicles on the road inside the target highway tunnel is read by the monitoring equipment inside the tunnel. The monitoring video data is preprocessed using the Retinex algorithm to enhance low light and eliminate the visual impact caused by sudden changes in light inside and outside the tunnel. The Retinex algorithm is a well-known existing technology. Implementers can choose other feasible image enhancement algorithms. This embodiment does not limit this.

[0043] Secondly, the visibility inside the highway tunnel is measured in real time using the forward scattering method. A temperature and humidity sensor is installed at the center of the highway tunnel to collect the temperature and humidity data in real time. Visibility, temperature, and humidity are all collected synchronously. In this embodiment, the collection frequency is 10Hz, but implementers can set it according to actual conditions; this embodiment does not impose any restrictions on this.

[0044] S2. Based on the current temperature and humidity ratio inside the highway tunnel and the difference between the visibility and historical visibility, determine the environmental parameter superiority of the current highway tunnel.

[0045] This application aims to provide deceleration warnings for vehicles in a highway tunnel through an electromechanical system. Since vehicle performance in a tunnel is related to the tunnel environment and real-time vehicle conditions, this embodiment first evaluates the environmental parameter suitability of the current tunnel based on multiple environmental factors, specifically:

[0046] Since vehicle performance within a tunnel is influenced by multiple environmental parameters and complex vehicle conditions, and monitoring videos cannot adequately reflect these environmental factors, this embodiment uses collected environmental parameters to assist in the analysis and judgment of the monitoring video process. Tunnel temperature, humidity, and visibility directly affect vehicle performance and driver safety. Lower temperatures and higher humidity significantly reduce the road surface friction coefficient, potentially leading to icing and further increasing the risk of vehicle skidding, thus reducing handling performance. Conversely, lower visibility causes greater visual interference for the driver. Therefore, the combined impact of the environment on vehicle performance and driver interference is greater, resulting in a poorer overall environmental quality.

[0047] Based on the above analysis, this embodiment calculates the ring parameter dominance of the current highway tunnel. The specific calculation process is as follows:

[0048] In the formula, This represents the current environmental parameter optimization of the highway tunnel, where T is the current temperature inside the highway tunnel, C is the current humidity inside the highway tunnel, and A is the current visibility inside the highway tunnel. This is the average of all visibility data collected within the current preset historical time period. In this embodiment, the length of the historical time period is three months; however, implementers can set this value according to their actual circumstances, and this embodiment does not impose any restrictions on it.

[0049] Specifically, for ease of calculation, the average of all visibility data collected within the current preset historical time period is used. When the value is zero, remove one maximum and one minimum value and recalculate until the value is not zero.

[0050] It should be understood that if the temperature and humidity ratio inside a highway tunnel... The larger the value, the better the friction performance of the tunnel surface, and the less impact it has on the vehicle's handling performance. At the same time, the clearer the visibility in the tunnel compared to historical levels, the less interference it has on the driver's vision. This further indicates that the real-time environmental parameters in the tunnel are excellent, and the greater the environmental parameter optimization.

[0051] S3 uses a neural network model to extract and track vehicles in the surveillance video of a highway tunnel; it analyzes the width difference between each vehicle and its neighboring vehicles, as well as the lateral distance between the vehicle centers, to determine the degree of visual obstruction interference for each vehicle.

[0052] This embodiment uses a neural network model to locate vehicles within the surveillance video of a highway tunnel. Specifically, the neural network model can select vehicles within the surveillance video using bounding boxes. The neural network model selected in this embodiment is YOLOv5. While neural network models for identifying vehicles within surveillance video are a well-known existing technology, implementers can choose other feasible existing neural network models; this embodiment does not impose any restrictions on this.

[0053] Because during vehicle movement, in a single lane, the wider the vehicle in front, the more likely it is to obstruct the view of vehicles behind. Furthermore, in the direction of traffic flow, if the vehicle in front is more to the left relative to the vehicle behind in the perpendicular direction of traffic flow, and is also relatively wider, it will cause stronger obstruction of the view for the driver of the following vehicle. Therefore, this embodiment determines the degree of obstruction of view for each vehicle based on the width difference between each vehicle and its neighboring vehicle in front, as well as the lateral distance between the centers of the vehicles. Specifically:

[0054] Since the neural network model uses bounding boxes to select vehicles in the surveillance video, this embodiment uses the width of the bounding box to represent the width of the vehicle. Furthermore, the center point of the bounding box is recorded as the center point of the vehicle in the surveillance video. The field-of-view occlusion interference degree of each vehicle in the current tunnel surveillance video is calculated using the following expression:

[0055] In the formula, norm() is the normalization function, and Q represents the field-of-view occlusion interference degree of each vehicle in the current tunnel monitoring video. This represents the width of the adjacent vehicle in front of each vehicle in the current tunnel surveillance video. This represents the width of each vehicle in the current tunnel surveillance video. The lateral distance between each vehicle and its adjacent vehicle in the current tunnel surveillance video is denoted as the lateral offset displacement. A schematic diagram of the lateral offset displacement is shown below. Figure 2 As shown.

[0056] It should be noted that when all vehicles in the surveillance video are aligned with the center point of the current vehicle, the lateral spacing Y is set to 0. In the surveillance video, the vertical direction represents the vehicle's driving direction, and the horizontal direction represents the direction perpendicular to the vehicle's driving direction.

[0057] If the width of each vehicle is relative to the width of the vehicle in front of it. The wider the width, the greater the lateral offset of each vehicle relative to its neighboring vehicle in the direction perpendicular to the traffic flow. The larger the value, the greater the obstruction of the vehicle's view by the vehicle in front, and the more necessary it is to issue a speed reduction warning to that vehicle.

[0058] S4. Based on the speed and acceleration differences between each vehicle and its neighboring vehicles, determine the rear-end collision risk level of each vehicle; based on the length differences between each vehicle and its neighboring vehicles, and in combination with the ring parameter dominance and the rear-end collision risk level, determine the speed acceleration coefficient of each vehicle.

[0059] Rear-end collisions are a common accident in highway tunnels. The faster the following vehicle is traveling while the vehicle in front is traveling slower, the more dangerous the situation becomes. Therefore, vehicle speed is a necessary factor to consider when issuing warnings for highway tunnels. In addition, even when the speeds of the vehicles in front and behind are equal, there is a possibility that the following vehicle is accelerating while the vehicle in front is decelerating. Although the speeds are equal in this situation, it is temporary, and a rear-end collision is still possible. Therefore, the impact of vehicle acceleration on the risk of a rear-end collision must also be considered.

[0060] Based on the above analysis, this embodiment identifies vehicles in the surveillance video using a neural network model, tracks the vehicles to calculate their speed, and simultaneously calculates their acceleration based on the speed. The calculation of vehicle speed and acceleration are existing, well-known techniques, and will not be described in detail here.

[0061] Therefore, the rear-end collision risk of each vehicle in the current tunnel surveillance video is calculated using the following expression:

[0062] In the formula, This indicates the rear-end collision risk level of each vehicle in the current tunnel surveillance video. This represents the acceleration of each vehicle in the current tunnel surveillance video. This refers to the acceleration of the nearest vehicle in the current tunnel surveillance video. The speed of each vehicle in the current tunnel surveillance video. Let `norm()` be the speed of the nearest vehicle in the current tunnel surveillance video, and `norm()` be the normalization function. This is denoted as the first ratio.

[0063] It should be noted that, to ensure the calculation results are meaningful, in this embodiment of the application, when performing fractional operations, if the denominator is 0, a parameter adjustment factor can be added to the denominator to prevent the denominator from being 0. This parameter adjustment factor is a very small positive number. For example, the value of this parameter adjustment factor can be 0.01. Its specific value can be set by the implementer according to the actual situation, and this embodiment of the application does not impose a specific limitation.

[0064] Specifically, the normalization (or normalization function) in the embodiments of this application is specifically minimum-maximum normalization.

[0065] It should be understood that due to the relatively enclosed space and short visibility within tunnels, traffic flow is somewhat restricted. When the acceleration and speed of a following vehicle are high, its reaction time becomes shorter, making it easier to be unable to adjust its speed or brake in time. Especially when the vehicle in front suddenly decelerates or stops, the probability of a rear-end collision increases significantly, thus increasing the risk of rear-end collisions.

[0066] Secondly, the risk of rear-end collisions between adjacent vehicles in a tunnel is not only reflected by vehicle speed. The smaller the real-time distance between the center points of the two vehicles, the greater the risk of a rear-end collision. Considering the differences in size between different vehicles in real-world scenarios, for example, a large truck is significantly longer than a small car. Therefore, for longer vehicles, the actual distance between them and adjacent vehicles is shorter than the distance reflected by the center point. At the same time, the higher the risk of a rear-end collision and the lower the real-time tunnel's environmental parameter optimization, the greater the impact on vehicle driving and the easier it is to cause a vehicle accident. Therefore, timely vehicle warnings are necessary.

[0067] Based on the above analysis, this embodiment calculates the speed acceleration coefficient of each vehicle in the current tunnel monitoring video, specifically as follows:

[0068] First, determine the longitudinal distance between the center points of each vehicle and its adjacent preceding vehicle in the surveillance video. Calculate the average length of each vehicle and its adjacent preceding vehicle in the surveillance video, denoted as the first average. In this embodiment, the vehicle length is represented by the length of the frame containing the vehicle in the surveillance video.

[0069] In the formula, This represents the speed acceleration coefficient for each vehicle in the current tunnel surveillance video. This is the average length of each vehicle in the current tunnel surveillance video and its adjacent vehicle in front. Let U represent the longitudinal distance between the center points of each vehicle and its adjacent preceding vehicle in the current tunnel monitoring video, and let U represent the circumferential parameter optimization of the current highway tunnel. This indicates the rear-end collision risk level of each vehicle in the current tunnel surveillance video. This is denoted as the second ratio.

[0070] It should be understood that the larger the length of the vehicle in the tunnel compared to the vehicle in front, and the greater the distance between the center positioning coordinates of the two vehicles... The smaller the value, the more dangerous the distance between the analyzed vehicle and the vehicle in front, and the higher the risk of a rear-end collision. The higher the value, the worse the real-time environmental performance inside the tunnel, which further indicates that the current vehicle speed trend is too high, and the more necessary it is to issue a deceleration warning.

[0071] S5, by combining the vehicle speed acceleration coefficient and the field of vision obstruction interference degree, the deceleration necessity of each vehicle is obtained, and by combining the density of vehicles in front of each vehicle in the tunnel, the deceleration urgency of each vehicle is determined.

[0072] Furthermore, if the speed acceleration coefficient of each vehicle in the current tunnel monitoring video is larger in the direction of traffic flow, and the field of view obstruction interference in the direction perpendicular to traffic flow is higher, it further indicates that the real-time deceleration necessity of the vehicle is higher. Therefore, the deceleration necessity of each vehicle in the current tunnel monitoring video is calculated using the following expression:

[0073] In the formula, To determine the necessary deceleration levels for each vehicle in the current tunnel surveillance video, Q represents the speed acceleration coefficient of each vehicle in the current tunnel surveillance video, and Q represents the field-of-view obstruction interference of each vehicle in the current tunnel surveillance video.

[0074] Secondly, for vehicles inside the tunnel, speed control needs to be determined not only by considering the driving performance of nearby vehicles but also by taking into account the real-time road congestion and vehicle density within the tunnel. If, for any given vehicle, the more concentrated the traffic is on the road segment not yet traversed by that vehicle, the more congested the traffic situation that vehicle is about to encounter, requiring speed control in advance. Therefore, based on the necessity of deceleration, the urgency of deceleration for each vehicle inside the tunnel is determined by analyzing the road congestion situation on the road segment ahead of the tunnel. Specifically:

[0075] First, for each vehicle in the current tunnel surveillance video, count its distance from the tunnel exit, denoted as the untraveled distance within the tunnel. Simultaneously, count the number of vehicles within the untraveled distance of each vehicle in the surveillance video. Combining this with the deceleration necessity of each vehicle, calculate the deceleration urgency of each vehicle in the surveillance video. The specific expression is as follows:

[0076] In the formula, This indicates the urgency of deceleration for each vehicle in the current tunnel surveillance video. This represents the number of vehicles within the tunnel's untraveled distance as shown in the current tunnel surveillance video. This represents the distance each vehicle has not traveled within the tunnel as shown in the current tunnel surveillance video. This determines the necessary deceleration level for each vehicle in the current tunnel surveillance video. This is denoted as the third ratio.

[0077] The greater the total number of vehicles within the tunnel's untraveled distance, the more congested the traffic conditions the vehicle will face, and the higher the necessity for the vehicle to decelerate. This further indicates that the vehicle urgently needs to decelerate in order to control its speed, hence the greater the urgency of deceleration.

[0078] S6. Extract the faulty vehicle from the monitoring video of the highway tunnel, analyze the measured distance between each vehicle and the faulty vehicle in the tunnel, as well as the interval between vehicles, and determine the deceleration warning value for each vehicle in the tunnel based on the aforementioned deceleration urgency, and issue a deceleration warning to each vehicle in the current tunnel.

[0079] Furthermore, considering the possibility of sudden special vehicle conditions within highway tunnels, such as abnormal situations like stopped vehicles, vehicles driving in the wrong direction, or vehicles using hazard lights detected by surveillance video, while dispatching backup vehicles, timely warnings should be issued to vehicles behind these vehicles within the tunnel. This helps them to be aware of the sudden road conditions, extending their reaction time and enabling them to react promptly. Therefore, this embodiment adjusts the deceleration urgency values ​​for each vehicle based on special vehicle conditions to determine the deceleration warning values ​​for each vehicle within the tunnel, specifically as follows:

[0080] First, the YOLOv5 neural network model is used to mark vehicles with malfunctions in the tunnel monitoring video, locate the coordinates of the malfunctioning vehicles, and calculate the metric distance between the coordinates of each vehicle in the current tunnel monitoring video and the malfunctioning vehicle. Simultaneously, the number of vehicles in between each vehicle and the malfunctioning vehicle in the current tunnel monitoring video is extracted. Based on this, the deceleration warning value for each vehicle in the current tunnel is determined. The specific expression is as follows:

[0081] In the formula, M represents the deceleration warning value for each vehicle in the current tunnel monitoring video, and M represents the number of vehicles between each vehicle and the disabled vehicle in the current tunnel monitoring video. The distance between each vehicle and the faulty vehicle in the current tunnel monitoring video is measured. In this embodiment, the distance is calculated using Euclidean distance. Implementers can choose other existing feasible distance calculation methods, and this embodiment does not impose any restrictions on this. This indicates the urgency of deceleration for each vehicle in the current tunnel monitoring video, where norm() is the normalization function. The flowchart for determining the deceleration warning value for vehicles in the tunnel is shown below. Figure 3 As shown.

[0082] It should be noted that if there is no faulty vehicle in the current monitoring video tunnel, or if there is a faulty vehicle but it is located behind the driving vehicle, the impact of the faulty vehicle on the driving vehicle is relatively small. In this embodiment, the deceleration urgency of the corresponding driving vehicle is used as the deceleration warning value.

[0083] It should be understood that the closer the distance between the currently analyzed vehicle and the faulty vehicle, and the more vehicles are between the two vehicles, the greater the difficulty for the currently analyzed vehicle to control its speed in the face of subsequent emergencies, and the greater the urgency for it to decelerate. This further indicates that the speed of the currently analyzed vehicle needs to be decelerated and warned.

[0084] Furthermore, the average deceleration warning value of all vehicles passing through the tunnel within the current preset historical time period is calculated and denoted as the second average value. The normalized value of the difference between the deceleration warning value of each vehicle in the current tunnel monitoring video and the second average value is calculated. The larger the normalized value, the larger the deceleration warning value of the corresponding vehicle compared with the historical vehicles, the more complex the driving environment of the vehicle, and the more necessary it is to issue a deceleration warning for the vehicle.

[0085] Therefore, if the normalized value of a vehicle in the current tunnel monitoring video is greater than a preset threshold, the corresponding vehicle will be designated as a vehicle requiring warning. The highway tunnel electromechanical system will then issue a deceleration warning signal to the driver via the vehicle's GPS navigation system. Otherwise, the corresponding vehicle will not be designated as a vehicle requiring warning. In this embodiment, the preset threshold is set to 0.8. Implementers can set it according to actual conditions, and this embodiment does not impose any restrictions on it.

[0086] Based on the same inventive concept as the above method, this application also provides a highway tunnel video monitoring system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described highway tunnel video monitoring methods.

[0087] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0088] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0089] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A method for monitoring a highway tunnel with video, characterized in that, The method comprises the following steps: Real-time acquisition of temperature, humidity, and visibility in the highway tunnel, and monitoring video in the highway tunnel; Determination of the ring parameter setting degree of the current highway tunnel based on the temperature-humidity ratio in the current highway tunnel and the difference between the visibility and the historical visibility; Extraction and tracking of each vehicle in the monitoring video of the highway tunnel by a neural network model; analysis of the width difference between each vehicle and its adjacent vehicle and the lateral spacing of the vehicle center to determine the visual field covering interference degree of each vehicle; Determination of the rear-end risk degree of each vehicle based on the speed difference between each vehicle and its adjacent vehicle and the acceleration difference; determination of the speed-up coefficient of each vehicle based on the length difference between each vehicle and its adjacent vehicle, in combination with the ring parameter setting degree and the rear-end risk degree; Fusion of the speed-up coefficient and the visual field covering interference degree to obtain the deceleration necessity degree of each vehicle, and determination of the deceleration urgency degree of each vehicle in combination with the vehicle density in front of each vehicle in the tunnel; Extraction of the fault vehicle in the monitoring video of the highway tunnel, analysis of the metric distance between each vehicle in the tunnel and the fault vehicle and the interval vehicle, determination of the deceleration warning value of each vehicle in the tunnel in combination with the deceleration urgency degree, and deceleration warning for each vehicle in the current tunnel.

2. The highway tunnel video monitoring method of claim 1, wherein, The determination of the ring parameter setting degree comprises: Determination of the mean value of the visibility in the current preset historical time period in the tunnel, calculation of the difference between the current visibility in the tunnel and the mean value, and positive correlation of the ring parameter setting degree with the difference and the temperature-humidity ratio.

3. The highway tunnel video monitoring method of claim 1, wherein, The determination of the visual field covering interference degree comprises: Calculation of the ratio of the width of each vehicle to its adjacent front vehicle, and the visual field covering interference degree is the product of the ratio and the lateral spacing.

4. The highway tunnel video monitoring method of claim 1, wherein, The determination of the rear-end risk degree comprises: Calculation of the ratio of the speed of each vehicle to its adjacent front vehicle, denoted as a first ratio, calculation of the normalized result of the difference between the acceleration of each vehicle and its adjacent front vehicle, and the rear-end risk degree is the product of the first ratio and the normalized result.

5. The highway tunnel video monitoring method of claim 1, wherein, The determination of the speed-up coefficient comprises: Calculation of the mean value of the vehicle length of each vehicle and its adjacent front vehicle, denoted as a first mean value, determination of the longitudinal spacing between the center point of each vehicle and the center point of its adjacent front vehicle, calculation of the ratio of the first mean value to the longitudinal spacing, denoted as a second ratio; Determination of the reciprocal of the ring parameter setting degree, and the speed-up coefficient is the normalized result of the product of the second ratio, the reciprocal, and the rear-end risk degree.

6. The highway tunnel video monitoring method of claim 1, wherein, The deceleration necessity degree is the square root result of the sum of the square of the speed-up coefficient and the square of the visual field covering interference degree.

7. The highway tunnel video monitoring method of claim 1, wherein, The determination of the deceleration urgency degree comprises: Calculation of the ratio of the number of front vehicles of each vehicle in the tunnel to the distance from the tunnel exit, denoted as a third ratio, and the deceleration urgency degree is the product of the third ratio and the deceleration necessity degree.

8. The highway tunnel video monitoring method of claim 1, wherein, The determination of the deceleration warning value comprises: Calculation of the normalized value of the number of interval vehicles and the reciprocal of the metric distance, and positive correlation of the deceleration warning value with the normalized value and the deceleration urgency degree.

9. The highway tunnel video monitoring method of claim 1, wherein, The deceleration warning for each vehicle in the current tunnel comprises: The normalized value of the difference between the deceleration warning value of each vehicle in the current tunnel and the average of the deceleration warning values of all vehicles in the tunnel in the preset historical time period is calculated, and if the normalized value is greater than a preset threshold, the corresponding vehicle is regarded as a vehicle needing warning, otherwise, the corresponding vehicle is not regarded as a vehicle needing warning.

10. A highway tunnel video monitoring system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, The processor implements the steps of the method of any one of claims 1-9 when executing the computer program.

Images