Real-time monitoring system based on cloud computing

Abstract

The application discloses a real-time monitoring system operation method based on cloud computing, and the application uploads real-time road videos of forbidden directions and passing directions to a cloud platform through a camera equipment, extracts road historical data and traffic signal rules of the cloud platform, performs three-dimensional frame drawing on road pedestrians and vehicles based on the real-time road videos, predicts target directions of the pedestrians and vehicles according to lane marks, constructs a dynamic traffic flow direction graph, calculates straight light and turning light conflict coefficients based on the dynamic traffic flow direction graph, analyzes road passing capacity, adjusts traffic signal strategies, divides time periods according to time sequences based on the road traffic characteristics, formulates intelligent passing decision rules according to traffic conditions of different time periods, and automatically adjusts signal timing schemes according to real-time monitoring results in the execution process of the intelligent communication decision rules, so that the application solves the problems of low traffic road use efficiency and red light and green light waiting.

Description

Technical Field

[0001] This invention relates to the field of real-time monitoring technology, specifically to a cloud computing-based real-time monitoring system. Background Technology

[0002] With the rapid development of the social economy and the acceleration of urbanization, traffic problems are becoming increasingly prominent. The existing real-time monitoring resources of traffic light sections are not being fully utilized. Traditional traffic light control often uses fixed time intervals for signal switching, which cannot be flexibly adjusted according to real-time traffic conditions and pedestrian flow. This not only leads to low traffic efficiency but also increases vehicle waiting time and energy consumption, and also affects the pedestrian experience. Especially at night or during off-peak hours, when there are almost no vehicles or pedestrians on one side of the road, while people are waiting to cross on the other side, the fixed-time traffic light system obviously fails to effectively utilize road resources, often resulting in pedestrians waiting idly. In addition, some traffic light control on small road sections reduces pedestrian waiting time by allowing straight and turning vehicles to pass simultaneously, but this makes intersection passage complex and dangerous during peak traffic hours. Therefore, it is necessary to design a cloud-based real-time monitoring system to improve road use efficiency and optimize the pedestrian experience. Summary of the Invention

[0003] The purpose of this invention is to provide a cloud computing-based real-time monitoring system to solve the problems mentioned in the background section.

[0004] To address the aforementioned technical problems, the present invention provides the following technical solution: a cloud computing-based real-time monitoring system, comprising executing a cloud computing-based real-time monitoring method, the method including:

[0005] Step S1: Collect real-time road videos of prohibited and permitted directions using camera equipment and upload them to the cloud platform; extract historical road data and traffic signal rules from the cloud platform.

[0006] Step S2: Based on the real-time road video, create 3D bounding boxes for pedestrians and vehicles in the road, predict the target directions of pedestrians and vehicles according to lane markings, and construct a dynamic traffic flow map;

[0007] Step S3: Based on the dynamic traffic flow map, calculate the conflict coefficient of straight-ahead lights and turn signals, analyze road capacity, and adjust traffic signal strategies;

[0008] Step S4: Based on the road traffic characteristics, divide the time period according to the time sequence, and formulate intelligent traffic decision rules according to the traffic conditions in different time periods;

[0009] Step S5: During the execution of the intelligent communication decision rules, the signal timing scheme is automatically adjusted based on the real-time monitoring results.

[0010] According to the above technical solution, step S1 further includes the following steps:

[0011] Step S11: After performing preliminary preprocessing on the real-time road video at the local edge computing node, upload it to the cloud platform and extract the road layout from the real-time road video;

[0012] Step S12: Preset data storage period, set a reasonable retention time for the real-time road video according to requirements, store the real-time road video in chronological order, and add timestamp and location information to each video data;

[0013] Step S13: Mark the real-time channel video according to special dates such as rest days and holidays in the calendar. Weekdays are marked in green, ordinary rest days are marked in yellow, and holidays are marked in red. Regularly review the cloud platform storage and implement a periodic data deletion strategy.

[0014] According to the above technical solution, step S2 further includes the following steps:

[0015] Step S21: Based on the road layout, identify lane lines, divide the road laterally according to the lane lines, locate pedestrians and vehicles in the real-time road video, and label pedestrians and vehicles in the form of 3D bounding boxes. For motor vehicle lanes, mark the target direction of the 3D bounding box of motor vehicles according to the lane marking and vehicle position of each lane. Among them, the target direction of motor vehicles waiting for traffic lights is only determined as the left turn direction and the forward direction. The target direction is marked in green. The lane with the 3D bounding box of motor vehicles is set to 1, and the lane without the 3D bounding box of motor vehicles is set to 0. When a vehicle crosses two lanes, the lane with the larger position occupied by the 3D bounding box of motor vehicles is the target lane of the vehicle. Count the number of vehicles in the left turn direction and the number of vehicles in the forward direction for each lane.

[0016] Step S22: Create a 3D bounding box for non-motorized vehicles and pedestrians in the non-motorized vehicle lane. Predict the number of non-motorized vehicles and the target direction of pedestrians based on the vehicle turning direction and traffic light status. Set a threshold to determine the target direction. When the current prediction accuracy is greater than the threshold, mark the target direction in green. When the prediction accuracy is less than the threshold, mark the target direction in yellow. Count the number of vehicles turning left and non-motorized vehicles and pedestrians going forward. Analyze the probability of turning left and going straight based on the road history data. Estimate the number of non-motorized vehicles and pedestrians marked with yellow target directions in the left and forward directions.

[0017] Step S23: Based on the following conditions, when the number of 3D boxes in the left-turn direction is not 0, the traffic signal left-turn light for the corresponding lane is 1; when the number of 3D boxes in the left-turn direction is 0, the traffic signal left-turn light for the corresponding lane is 0; when the number of 3D boxes in the straight-ahead direction is not 0, the traffic signal straight-ahead light for the corresponding lane is 1; when the number of 3D boxes in the straight-ahead direction is 0, the traffic signal straight-ahead light for the corresponding lane is 0, construct the dynamic traffic flow map of the current road.

[0018] According to the above technical solution, step S3 further includes the following steps:

[0019] Step S31: Based on the dynamic traffic flow diagram, when the left turn signal and the straight signal are both 1, mark the number of three-dimensional boxes of motor vehicle lanes corresponding to the traffic lights in the target direction as green and the number of three-dimensional boxes of non-motor vehicle lanes as yellow, and mark the traffic lights in each direction of the current road in the same way.

[0020] Step S32: Based on the traffic signal rules, determine whether the original traffic signal rules allow straight-ahead and turning vehicles to pass simultaneously. If the determination is no, proceed to step S4. If the determination is yes, calculate the conflict coefficient of the straight-ahead and turning lights based on the number of left-turn and straight-ahead lights marked in the dynamic traffic flow diagram. The formula for calculating the conflict coefficient of the straight-ahead and turning lights is:

[0021] CI=(α*ML+FN*β*NL) / (α*(ML+MS)+FN*β*(NL+NS)

[0022] In the formula, CI represents the conflict coefficient of the straight-ahead light and the turn signal, α and β represent the conflict weights of the motor vehicle lane and the non-motor vehicle lane, respectively, ML represents the number of motor vehicles turning left, FN represents the non-motor vehicle flexibility factor, NL represents the predicted number of non-motor vehicles turning left, MS represents the number of motor vehicles going straight, and NS represents the predicted number of pedestrians and non-motor vehicles going straight.

[0023] Step S33: Set a threshold based on the conflict coefficient of the straight-ahead light and the turn signal. When the conflict coefficient of the straight-ahead light and the turn signal is greater than the threshold, vehicles going straight and turning are prohibited from passing at the same time. When the conflict coefficient of the straight-ahead light and the turn signal is less than the threshold, vehicles going straight and turning are allowed to pass at the same time. Adjust the time period according to the different colors of the calendar markers.

[0024] According to the above technical solution, step S4 further includes the following steps:

[0025] Step S41: Based on the traffic signal left turn light and the traffic signal straight light markings and the corresponding number of three-dimensional boxes, calculate the ratio between the current number of three-dimensional boxes for the prohibited direction and the passing direction. When the ratio is greater than a threshold, increase the green light timing for the prohibited direction. When the ratio is less than 20% of the threshold, follow the original traffic light timing. When the ratio is less than 20% of the threshold, increase the green light timing for the passing direction. Adjust the time period according to the different colors of the calendar markings.

[0026] Step S42: Based on the historical road video and the adjusted time period recorded, extract the nodes and key features of the time period, predict the traffic flow of each time period in the future, provide the corresponding traffic light timing, and formulate appropriate intelligent traffic decision rules.

[0027] According to the above technical solution, step S5 further includes: identifying the traffic light markings based on the dynamic traffic flow map; when the traffic light marking for the direction of travel is 0, ending the green light for the current direction of travel and switching to the next traffic light marking 1 in the original traffic signal rule switching sequence; when all traffic lights at the intersection are marked 0, following the original traffic signal rules.

[0028] According to the above technical solution, the system includes a monitoring data acquisition module, a traffic flow map construction module, a conflict coefficient adjustment signal module, and an intelligent traffic decision module.

[0029] The monitoring data acquisition module is used to collect real-time road videos of prohibited and permitted directions through camera equipment and upload them to the cloud platform, and extract historical road data and traffic signal rules from the cloud platform;

[0030] The traffic flow map construction module is used to draw three-dimensional boxes of pedestrians and vehicles in the road based on the real-time road video, predict the target direction of pedestrians and vehicles according to lane markings, and construct a dynamic traffic flow map.

[0031] The conflict coefficient adjustment signal module is used to calculate the conflict coefficient of straight-ahead lights and turn signals based on the dynamic traffic flow map, analyze road capacity, and adjust traffic signal strategies.

[0032] The intelligent traffic decision module is used to divide time periods according to the time sequence based on the road traffic characteristics, formulate intelligent traffic decision rules according to the traffic conditions in different time periods, and automatically adjust the signal timing scheme according to the real-time monitoring results during the execution of the intelligent communication decision rules.

[0033] According to the above technical solution, the traffic flow map construction module includes a 3D bounding box annotation module, a target direction prediction module, and a traffic flow map generation module:

[0034] The three-dimensional bounding box annotation module is used to identify lane lines based on the road layout, divide the road laterally according to the lane lines, locate pedestrians and vehicles in the real-time road video, and annotate pedestrians and vehicles in the form of three-dimensional bounding boxes, and perform three-dimensional bounding boxes on non-motorized vehicles and pedestrians in non-motorized lanes.

[0035] The target direction prediction module is used to mark the target direction of a vehicle's 3D bounding box based on the lane markings and vehicle positions of each lane. For vehicles waiting at traffic lights, the target direction is determined only as a left turn or forward direction, and is marked in green. Lanes with the vehicle's 3D bounding box are set to 1, and lanes without it are set to 0. When a vehicle crosses two lanes, the lane with the larger lane position occupied by the vehicle's 3D bounding box is identified as the vehicle's target lane. The module counts the number of vehicles waiting in each lane for both left turns and forward traffic. Based on vehicle deflection direction and traffic light status, it predicts the number of non-motorized vehicles and pedestrians' target directions. A threshold is set; if the current prediction accuracy is greater than the threshold, the target direction is marked in green; if the prediction accuracy is less than the threshold, the target direction is marked in yellow. The module counts the number of vehicles turning left and the number of non-motorized vehicles and pedestrians forward, and analyzes the probability of left turns and straight-ahead directions based on historical road data. It then estimates the number of non-motorized vehicles and pedestrians with yellow-marked target directions for both left turns and forward traffic.

[0036] The traffic flow map generation module is used to construct the dynamic traffic flow map of the current road based on the following conditions: when the number of three-dimensional boxes in the left-turn direction is not 0, the traffic signal left-turn light for that lane is 1; when the number of three-dimensional boxes in the left-turn direction is 0, the traffic signal left-turn light for that lane is 0; when the number of three-dimensional boxes in the straight-ahead direction is not 0, the traffic signal straight-ahead light for that lane is 1; when the number of three-dimensional boxes in the straight-ahead direction is 0, the traffic signal straight-ahead light for that lane is 0.

[0037] According to the above technical solution, the conflict coefficient adjustment signal module includes a traffic light marking module, a conflict coefficient calculation module, and a traffic signal adjustment module:

[0038] The traffic signal marking module is used to mark the number of three-dimensional boxes of motor vehicle lanes and the number of three-dimensional boxes of non-motor vehicle lanes corresponding to the traffic signal lights corresponding to the target direction as green and yellow, respectively, when the traffic signal left turn light and the traffic signal straight light are 1 in the dynamic traffic flow diagram.

[0039] The conflict coefficient calculation module is used to determine, based on the traffic signal rules, whether the original traffic signal rules allow straight-ahead and turning vehicles to pass simultaneously. If the determination is no, it jumps to step S4; if the determination is yes, it calculates the conflict coefficient of the straight-ahead and turning lights based on the number markings of the left-turn and straight-ahead lights in the dynamic traffic flow diagram. The formula for calculating the conflict coefficient of the straight-ahead and turning lights is as follows:

[0040] CI=(α*ML+FN*β*NL) / (α*(ML+MS)+FN*β*(NL+NS)

[0041] In the formula, CI represents the conflict coefficient of the straight-ahead light and the turn signal, α and β represent the conflict weights of the motor vehicle lane and the non-motor vehicle lane, respectively, ML represents the number of motor vehicles turning left, FN represents the non-motor vehicle flexibility factor, NL represents the predicted number of non-motor vehicles turning left, MS represents the number of motor vehicles going straight, and NS represents the predicted number of pedestrians and non-motor vehicles going straight.

[0042] The traffic signal adjustment module is used to set a threshold based on the conflict coefficient of the straight-ahead light and the turn signal. When the conflict coefficient of the straight-ahead light and the turn signal is greater than the threshold, vehicles going straight and turning are prohibited from passing at the same time. When the conflict coefficient of the straight-ahead light and the turn signal is less than the threshold, vehicles going straight and turning are allowed to pass at the same time. The adjustment time period is recorded according to the different colors of the calendar markers.

[0043] According to the above technical solution, the intelligent traffic decision-making module includes a signal timing adjustment module, a traffic flow prediction module, and a traffic light matching adjustment module:

[0044] The signal timing adjustment module is used to calculate the ratio between the current prohibited direction and the number of three-dimensional boxes based on the traffic signal left turn light and the traffic signal straight light markings and the corresponding number of three-dimensional boxes. When the ratio is greater than a threshold, the green light timing for the prohibited direction is increased. When the ratio is less than 20% of the threshold, the original traffic light timing is maintained. When the ratio is less than 20% of the threshold, the green light timing for the passing direction is increased. The time period is adjusted according to the different colors of the calendar markings.

[0045] The traffic flow prediction module is used to extract nodes and key features of the time period based on the historical road video and recorded adjustment time period, predict the traffic flow of each future time period, provide corresponding traffic light timing, and formulate appropriate intelligent traffic decision rules.

[0046] The traffic light matching and adjustment module is used to identify the markings of traffic lights based on the dynamic traffic flow map. When the marking of the traffic light in the direction of travel is 0, the green light for the current direction of travel ends and the signal jumps to the next traffic light marked 1 in the original traffic signal rule jump sequence. When the markings of all traffic lights in all directions at the intersection are 0, the original traffic lights are executed.

[0047] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: By collecting road videos and analyzing historical data in real time, the system can respond more quickly to changes in traffic conditions, improving the real-time performance and accuracy of traffic management. Utilizing three-dimensional frame-drawing technology and dynamic traffic flow maps, the system can more accurately predict and analyze potential traffic conflicts, adjust traffic signals, and reduce traffic accidents. By annotating traffic videos according to special dates in the calendar (such as rest days and holidays) and implementing corresponding data deletion strategies, the system can adapt to traffic demands at different times, optimize traffic flow, and reduce waiting times, thereby improving road utilization efficiency and optimizing the pedestrian experience. Attached Figure Description

[0048] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0049] Figure 1 This is a flowchart illustrating the execution of an intelligent traffic signal control method by a cloud computing-based real-time monitoring system according to Embodiment 1 of the present invention.

[0050] Figure 2 This is a schematic diagram of the module composition of the cloud computing-based real-time monitoring system provided in Embodiment 2 of the present invention. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Example 1: This example can be applied to intelligent traffic signal control scenarios. The method can be executed by the cloud-based real-time monitoring system provided in this example. Figure 1 The flowchart illustrating the execution of an intelligent traffic signal control method by a cloud-based real-time monitoring system according to Embodiment 1 of the present invention is shown below. The method specifically includes the following steps:

[0053] Step S1: Collect real-time road videos of prohibited and permitted directions using camera equipment and upload them to the cloud platform; extract historical road data and traffic signal rules from the cloud platform.

[0054] Step S2: Based on the real-time road video, create 3D bounding boxes for pedestrians and vehicles in the road, predict the target directions of pedestrians and vehicles according to lane markings, and construct a dynamic traffic flow map;

[0055] Step S3: Based on the dynamic traffic flow map, calculate the conflict coefficient of straight-ahead lights and turn signals, analyze road capacity, and adjust traffic signal strategies;

[0056] Step S4: Based on the road traffic characteristics, divide the time period according to the time sequence, and formulate intelligent traffic decision rules according to the traffic conditions in different time periods;

[0057] Step S5: During the execution of the intelligent communication decision rules, the signal timing scheme is automatically adjusted based on the real-time monitoring results.

[0058] In this embodiment of the invention, step S1 further includes the following steps:

[0059] Step S11: After performing preliminary preprocessing on the real-time road video at the local edge computing node, upload it to the cloud platform and extract the road layout from the real-time road video;

[0060] Step S12: Preset data storage period, set a reasonable retention time for the real-time road video according to requirements, store the real-time road video in chronological order, and add timestamp and location information to each video data;

[0061] Step S13: Mark the real-time channel video according to special dates such as rest days and holidays in the calendar. Weekdays are marked in green, ordinary rest days are marked in yellow, and holidays are marked in red. Regularly review the cloud platform storage and implement a periodic data deletion strategy.

[0062] In this embodiment of the invention, step S2 further includes the following steps:

[0063] Step S21: Based on the road layout, identify lane lines, divide the road laterally according to the lane lines, locate pedestrians and vehicles in the real-time road video, and label pedestrians and vehicles in the form of 3D bounding boxes. For motor vehicle lanes, mark the target direction of the 3D bounding box of motor vehicles according to the lane marking and vehicle position of each lane. Among them, the target direction of motor vehicles waiting for traffic lights is only determined as the left turn direction and the forward direction. The target direction is marked in green. The lane with the 3D bounding box of motor vehicles is set to 1, and the lane without the 3D bounding box of motor vehicles is set to 0. When a vehicle crosses two lanes, the lane with the larger position occupied by the 3D bounding box of motor vehicles is the target lane of the vehicle. Count the number of vehicles in the left turn direction and the number of vehicles in the forward direction for each lane.

[0064] Step S22: Create a 3D bounding box for non-motorized vehicles and pedestrians in the non-motorized vehicle lane. Predict the number of non-motorized vehicles and the target direction of pedestrians based on the vehicle turning direction and traffic light status. Set a threshold to determine the target direction. When the current prediction accuracy is greater than the threshold, mark the target direction in green. When the prediction accuracy is less than the threshold, mark the target direction in yellow. Count the number of vehicles turning left and non-motorized vehicles and pedestrians going forward. Analyze the probability of turning left and going straight based on the road history data. Estimate the number of non-motorized vehicles and pedestrians marked with yellow target directions in the left and forward directions.

[0065] Step S23: Based on the following conditions, when the number of 3D boxes in the left-turn direction is not 0, the traffic signal left-turn light for the corresponding lane is 1; when the number of 3D boxes in the left-turn direction is 0, the traffic signal left-turn light for the corresponding lane is 0; when the number of 3D boxes in the straight-ahead direction is not 0, the traffic signal straight-ahead light for the corresponding lane is 1; when the number of 3D boxes in the straight-ahead direction is 0, the traffic signal straight-ahead light for the corresponding lane is 0, construct the dynamic traffic flow map of the current road.

[0066] In this embodiment of the invention, step S3 further includes the following steps:

[0067] Step S31: Based on the dynamic traffic flow diagram, when the left turn signal and the straight signal are both 1, mark the number of three-dimensional boxes of motor vehicle lanes corresponding to the traffic lights in the target direction as green and the number of three-dimensional boxes of non-motor vehicle lanes as yellow, and mark the traffic lights in each direction of the current road in the same way.

[0068] Step S32: Based on the traffic signal rules, determine whether the original traffic signal rules allow straight-ahead and turning vehicles to pass simultaneously. If the determination is no, proceed to step S4. If the determination is yes, calculate the conflict coefficient of the straight-ahead and turning lights based on the number of left-turn and straight-ahead lights marked in the dynamic traffic flow diagram. The formula for calculating the conflict coefficient of the straight-ahead and turning lights is:

[0069] CI=(α*ML+FN*β*NL) / (α*(ML+MS)+FN*β*(NL+NS)

[0070] In the formula, CI represents the conflict coefficient of the straight-ahead light and the turn signal, α and β represent the conflict weights of the motor vehicle lane and the non-motor vehicle lane, respectively, ML represents the number of motor vehicles turning left, FN represents the non-motor vehicle flexibility factor, NL represents the predicted number of non-motor vehicles turning left, MS represents the number of motor vehicles going straight, and NS represents the predicted number of pedestrians and non-motor vehicles going straight.

[0071] Step S33: Set a threshold based on the conflict coefficient of the straight-ahead light and the turn signal. When the conflict coefficient of the straight-ahead light and the turn signal is greater than the threshold, vehicles going straight and turning are prohibited from passing at the same time. When the conflict coefficient of the straight-ahead light and the turn signal is less than the threshold, vehicles going straight and turning are allowed to pass at the same time. Adjust the time period according to the different colors of the calendar markers.

[0072] In this embodiment of the invention, step S4 further includes the following steps:

[0073] Step S41: Based on the traffic signal left turn light and the traffic signal straight light markings and the corresponding number of three-dimensional boxes, calculate the ratio between the current number of three-dimensional boxes for the prohibited direction and the passing direction. When the ratio is greater than a threshold, increase the green light timing for the prohibited direction. When the ratio is less than 20% of the threshold, follow the original traffic light timing. When the ratio is less than 20% of the threshold, increase the green light timing for the passing direction. Adjust the time period according to the different colors of the calendar markings.

[0074] Step S42: Based on the historical road video and the adjusted time period recorded, extract the nodes and key features of the time period, predict the traffic flow of each time period in the future, provide the corresponding traffic light timing, and formulate appropriate intelligent traffic decision rules.

[0075] In this embodiment of the invention, step S5 further includes: identifying the markers of traffic lights based on the dynamic traffic flow map; when the marker of the traffic light in the direction of travel is 0, ending the green light in the current direction of travel and switching to the next traffic light marked 1 in the original traffic signal rule switching sequence; when the markers of all traffic lights in the intersection are 0, executing according to the original traffic lights.

[0076] Example 2: Example 2 of the present invention provides a real-time monitoring system based on cloud computing. Figure 2 This is a schematic diagram of the module composition of the cloud computing-based real-time monitoring system provided in Embodiment 2 of the present invention, as shown below. Figure 2 As shown, the system includes a monitoring data acquisition module, a traffic flow map construction module, a conflict coefficient adjustment signal module, and an intelligent traffic decision-making module.

[0077] The monitoring data acquisition module is used to collect real-time road videos of prohibited and permitted directions through camera equipment and upload them to the cloud platform, and extract historical road data and traffic signal rules from the cloud platform;

[0078] The traffic flow map construction module is used to draw three-dimensional boxes of pedestrians and vehicles in the road based on the real-time road video, predict the target direction of pedestrians and vehicles according to lane markings, and construct a dynamic traffic flow map.

[0079] The conflict coefficient adjustment signal module is used to calculate the conflict coefficient of straight-ahead lights and turn signals based on the dynamic traffic flow map, analyze road capacity, and adjust traffic signal strategies.

[0080] The intelligent traffic decision module is used to divide time periods according to the time sequence based on the road traffic characteristics, formulate intelligent traffic decision rules according to the traffic conditions in different time periods, and automatically adjust the signal timing scheme according to the real-time monitoring results during the execution of the intelligent communication decision rules.

[0081] In some embodiments of the present invention, the traffic flow map construction module includes a 3D bounding box annotation module, a target direction prediction module, and a traffic flow map generation module:

[0082] The three-dimensional bounding box annotation module is used to identify lane lines based on the road layout, divide the road laterally according to the lane lines, locate pedestrians and vehicles in the real-time road video, and annotate pedestrians and vehicles in the form of three-dimensional bounding boxes, and perform three-dimensional bounding boxes on non-motorized vehicles and pedestrians in non-motorized lanes.

[0083] The target direction prediction module is used to mark the target direction of a vehicle's 3D bounding box based on the lane markings and vehicle positions of each lane. For vehicles waiting at traffic lights, the target direction is determined only as a left turn or forward direction, and is marked in green. Lanes with the vehicle's 3D bounding box are set to 1, and lanes without it are set to 0. When a vehicle crosses two lanes, the lane with the larger lane position occupied by the vehicle's 3D bounding box is identified as the vehicle's target lane. The module counts the number of vehicles waiting in each lane for both left turns and forward traffic. Based on vehicle deflection direction and traffic light status, it predicts the number of non-motorized vehicles and pedestrians' target directions. A threshold is set; if the current prediction accuracy is greater than the threshold, the target direction is marked in green; if the prediction accuracy is less than the threshold, the target direction is marked in yellow. The module counts the number of vehicles turning left and the number of non-motorized vehicles and pedestrians forward, and analyzes the probability of left turns and straight-ahead directions based on historical road data. It then estimates the number of non-motorized vehicles and pedestrians with yellow-marked target directions for both left turns and forward traffic.

[0084] The traffic flow map generation module is used to construct the dynamic traffic flow map of the current road based on the following conditions: when the number of three-dimensional boxes in the left-turn direction is not 0, the traffic signal left-turn light for that lane is 1; when the number of three-dimensional boxes in the left-turn direction is 0, the traffic signal left-turn light for that lane is 0; when the number of three-dimensional boxes in the straight-ahead direction is not 0, the traffic signal straight-ahead light for that lane is 1; when the number of three-dimensional boxes in the straight-ahead direction is 0, the traffic signal straight-ahead light for that lane is 0.

[0085] In some embodiments of the present invention, the conflict coefficient adjustment signal module includes a traffic light marking module, a conflict coefficient calculation module, and a traffic signal adjustment module:

[0086] The traffic signal marking module is used to mark the number of three-dimensional boxes of motor vehicle lanes and the number of three-dimensional boxes of non-motor vehicle lanes corresponding to the traffic signal lights corresponding to the target direction as green and yellow, respectively, when the traffic signal left turn light and the traffic signal straight light are 1 in the dynamic traffic flow diagram.

[0087] The conflict coefficient calculation module is used to determine, based on the traffic signal rules, whether the original traffic signal rules allow straight-ahead and turning vehicles to pass simultaneously. If the determination is no, it jumps to step S4; if the determination is yes, it calculates the conflict coefficient of the straight-ahead and turning lights based on the number markings of the left-turn and straight-ahead lights in the dynamic traffic flow diagram. The formula for calculating the conflict coefficient of the straight-ahead and turning lights is as follows:

[0088] CI=(α*ML+FN*β*NL) / (α*(ML+MS)+FN*β*(NL+NS)

[0089] In the formula, CI represents the conflict coefficient of the straight-ahead light and the turn signal, α and β represent the conflict weights of the motor vehicle lane and the non-motor vehicle lane, respectively, ML represents the number of motor vehicles turning left, FN represents the non-motor vehicle flexibility factor, NL represents the predicted number of non-motor vehicles turning left, MS represents the number of motor vehicles going straight, and NS represents the predicted number of pedestrians and non-motor vehicles going straight.

[0090] The traffic signal adjustment module is used to set a threshold based on the conflict coefficient of the straight-ahead light and the turn signal. When the conflict coefficient of the straight-ahead light and the turn signal is greater than the threshold, vehicles going straight and turning are prohibited from passing at the same time. When the conflict coefficient of the straight-ahead light and the turn signal is less than the threshold, vehicles going straight and turning are allowed to pass at the same time. The adjustment time period is recorded according to the different colors of the calendar markers.

[0091] In some embodiments of the present invention, the intelligent traffic decision module includes a signal timing adjustment module, a traffic flow prediction module, and a traffic light matching adjustment module:

[0092] The signal timing adjustment module is used to calculate the ratio between the current prohibited direction and the number of three-dimensional boxes based on the traffic signal left turn light and the traffic signal straight light markings and the corresponding number of three-dimensional boxes. When the ratio is greater than a threshold, the green light timing for the prohibited direction is increased. When the ratio is less than 20% of the threshold, the original traffic light timing is maintained. When the ratio is less than 20% of the threshold, the green light timing for the passing direction is increased. The time period is adjusted according to the different colors of the calendar markings.

[0093] The traffic flow prediction module is used to extract nodes and key features of the time period based on the historical road video and recorded adjustment time period, predict the traffic flow of each future time period, provide corresponding traffic light timing, and formulate appropriate intelligent traffic decision rules.

[0094] The traffic light matching and adjustment module is used to identify the markings of traffic lights based on the dynamic traffic flow map. When the marking of the traffic light in the direction of travel is 0, the green light for the current direction of travel ends and the signal jumps to the next traffic light marked 1 in the original traffic signal rule jump sequence. When the markings of all traffic lights in all directions at the intersection are 0, the original traffic lights are executed.

[0095] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0096] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A cloud computing-based real-time monitoring system, characterized in that: The system includes a monitoring data acquisition module, a traffic flow map construction module, a conflict coefficient adjustment signal module, and an intelligent traffic decision-making module. The monitoring data acquisition module is used to collect real-time road videos of prohibited and permitted directions through camera equipment and upload them to the cloud platform, and extract historical road data and traffic signal rules from the cloud platform; The traffic flow map construction module is used to draw three-dimensional boxes of pedestrians and vehicles in the road based on the real-time road video, predict the target direction of pedestrians and vehicles according to lane markings, and construct a dynamic traffic flow map. The conflict coefficient adjustment signal module is used to calculate the conflict coefficient of straight-ahead lights and turn signals based on the dynamic traffic flow map, analyze road capacity, and adjust traffic signal strategies. The intelligent traffic decision module is used to divide time periods according to time sequence based on road traffic characteristics, formulate intelligent traffic decision rules according to traffic conditions in different time periods, and automatically adjust the signal timing scheme according to real-time monitoring results during the execution of the intelligent traffic decision rules. The operation method of the real-time monitoring system includes the following steps: Step S1: Collect real-time road videos of prohibited and permitted directions using camera equipment and upload them to the cloud platform; extract historical road data and traffic signal rules from the cloud platform. Step S2: Based on the real-time road video, create 3D bounding boxes for pedestrians and vehicles in the road, predict the target directions of pedestrians and vehicles according to lane markings, and construct a dynamic traffic flow map; Step S3: Based on the dynamic traffic flow map, calculate the conflict coefficient of straight-ahead lights and turn signals, analyze road capacity, and adjust traffic signal strategies; Step S4: Based on the road traffic characteristics, divide the time period according to the time sequence, and formulate intelligent traffic decision rules according to the traffic conditions in different time periods; Step S5: During the execution of the intelligent passage decision rules, the signal timing scheme is automatically adjusted based on real-time monitoring results; Step S3 further includes the following steps: Step S31: When the left turn signal and the straight signal in the dynamic traffic flow diagram are both 1, mark the number of three-dimensional boxes of the motor vehicle lane corresponding to the traffic signal corresponding to the target direction as green and the number of three-dimensional boxes of the non-motor vehicle lane as yellow. Mark the traffic signals in each direction of the current road in the same way. Step S32: Based on the traffic signal rules, determine whether the original traffic signal rules allow straight-ahead and turning vehicles to pass simultaneously. If the determination is no, proceed to step S4. If the determination is yes, calculate the conflict coefficient of the straight-ahead and turning lights according to the number markings of the left-turn and straight-ahead lights in the dynamic traffic flow diagram. The formula for calculating the conflict coefficient of the straight-ahead and turning lights is: In the formula, This indicates the conflict coefficient between the straight-ahead lights and the turn signals. and These represent the conflict weights for motor vehicle lanes and non-motor vehicle lanes, respectively. Indicates the number of vehicles turning left. Indicates the flexibility factor for non-motorized vehicles. This indicates the predicted number of non-motorized vehicles turning left, and the predicted number of motorized vehicles going straight. This indicates the predicted number of pedestrians and non-motorized vehicles traveling straight. Step S33: Set a threshold based on the conflict coefficient of the straight-ahead light and the turn signal. When the conflict coefficient of the straight-ahead light and the turn signal is greater than the threshold, vehicles going straight and turning are prohibited from passing at the same time. When the conflict coefficient of the straight-ahead light and the turn signal is less than the threshold, vehicles going straight and turning are allowed to pass at the same time. Adjust the time period according to the different colors of the calendar markers.

2. The cloud computing-based real-time monitoring system according to claim 1, characterized in that: The traffic flow map construction module includes a 3D bounding box annotation module, a target direction prediction module, and a traffic flow map generation module. The three-dimensional bounding box annotation module is used to identify lane lines based on the road layout, divide the road laterally according to the lane lines, locate pedestrians and vehicles in the real-time road video, and annotate pedestrians and vehicles in the form of three-dimensional bounding boxes, and perform three-dimensional bounding boxes on non-motorized vehicles and pedestrians in non-motorized lanes. The target direction prediction module is used to mark the target direction of a vehicle's 3D bounding box based on the lane markings and vehicle positions of each lane in the motor vehicle lanes. For vehicles waiting at traffic lights, the target direction is determined only as either a left turn or a straight-ahead direction, and is marked in green. Lanes with the vehicle's 3D bounding box are set to 1, and lanes without it are set to 0. When a vehicle crosses two lanes, the lane with the larger occupied area of ​​the vehicle's 3D bounding box is identified as the vehicle's target lane. The module counts the number of vehicles waiting to turn left and the number of vehicles going straight in each lane. Based on vehicle deflection direction and traffic light status, it predicts the target directions of non-motorized vehicles and pedestrians. A threshold is set to determine the target direction; if the current prediction accuracy is greater than the threshold, it is marked in green; if the prediction accuracy is less than the threshold, it is marked in yellow. The module counts the number of vehicles turning left and the number of non-motorized vehicles and pedestrians going straight. Based on historical road data, it analyzes the probability of left turns and straight-ahead directions and estimates the number of non-motorized vehicles and pedestrians with yellow-marked target directions. The traffic flow map generation module is used to construct the dynamic traffic flow map of the current road based on the following conditions: when the number of 3D boxes in the left-turn direction is not 0, the traffic signal left-turn light for that lane is 1; when the number of 3D boxes in the left-turn direction is 0, the traffic signal left-turn light for that lane is 0; when the number of 3D boxes in the straight-ahead direction is not 0, the traffic signal straight-ahead light for that lane is 1; when the number of 3D boxes in the straight-ahead direction is 0, the traffic signal straight-ahead light for that lane is 0.

3. The cloud computing-based real-time monitoring system according to claim 2, characterized in that: The conflict coefficient adjustment signal module includes a traffic light marking module, a conflict coefficient calculation module, and a traffic signal adjustment module. The traffic signal marking module is used to mark the number of three-dimensional frames of motor vehicle lanes and the number of three-dimensional frames of non-motor vehicle lanes corresponding to the traffic signal lights corresponding to the target direction as green and yellow, respectively, when the traffic signal left turn light and the traffic signal straight light are 1 in the dynamic traffic flow diagram. The conflict coefficient calculation module is used to make a judgment based on the traffic signal rules to determine whether the original traffic signal rules allow straight-going and turning vehicles to pass at the same time. If the judgment is no, it jumps to step S4. If the judgment is yes, it calculates the conflict coefficient of the straight-going light and the turning light according to the number markings of the left-turn light and the straight-going light in the dynamic traffic flow diagram. The traffic signal adjustment module is used to set a threshold based on the conflict coefficient of the straight-ahead light and the turn signal. When the conflict coefficient of the straight-ahead light and the turn signal is greater than the threshold, vehicles going straight and turning are prohibited from passing at the same time. When the conflict coefficient of the straight-ahead light and the turn signal is less than the threshold, vehicles going straight and turning are allowed to pass at the same time. The adjustment time period is recorded according to the different colors of the calendar markers.

4. The cloud computing-based real-time monitoring system according to claim 3, characterized in that: The intelligent traffic decision-making module includes a signal timing adjustment module, a traffic flow prediction module, and a traffic light matching adjustment module. The signal timing adjustment module is used to calculate the ratio between the number of three-dimensional frames in the current prohibited direction and the number of three-dimensional frames in the traffic signal left turn light and the traffic signal straight light, as well as the corresponding number of three-dimensional frames. When the ratio is greater than a threshold, the green light timing for the prohibited direction is increased. When the ratio is less than 20% of the threshold, the original traffic light timing is maintained. When the ratio is less than 20% of the threshold, the green light timing for the passing direction is increased. The time period is adjusted according to the different colors of the calendar markers. The traffic flow prediction module is used to extract nodes and key features of the time period based on historical road videos and records, predict traffic flow in future time periods, provide corresponding traffic light timings, and formulate intelligent traffic decision rules. The traffic light matching and adjustment module is used to identify the markings of traffic lights based on the dynamic traffic flow map. When the marking of the traffic light in the direction of travel is 0, the green light for the current direction of travel ends and the signal jumps to the next traffic light with a marking of 1 in the original traffic signal rule jump sequence. When the markings of all traffic lights in all directions at the intersection are 0, the original traffic lights are executed.

5. The real-time monitoring system based on cloud computing according to claim 1, characterized in that: Step S1 further includes the following steps: Step S11: After performing preliminary preprocessing on the real-time road video at the local edge computing node, upload it to the cloud platform and extract the road layout from the real-time road video; Step S12: Preset data storage period, set a reasonable retention time for the real-time road video according to requirements, store the real-time road video in chronological order, and add timestamp and location information to each video data; Step S13: Mark the real-time road video according to the rest days and holidays in the calendar. Weekdays are marked in green, ordinary rest days are marked in yellow, and holidays are marked in red. The cloud platform storage is reviewed regularly, and a periodic data deletion strategy is implemented.

6. The real-time monitoring system based on cloud computing according to claim 5, characterized in that: Step S2 further includes the following steps: Step S21: Identify lane lines based on the road layout, divide the road laterally according to the lane lines, locate pedestrians and vehicles in the real-time road video, and label pedestrians and vehicles in the form of 3D bounding boxes. For motor vehicle lanes, mark the target direction of the 3D bounding box of motor vehicles according to the lane marking and vehicle position of each lane. Among them, the target direction of motor vehicles waiting for traffic lights is only determined as the left turn direction and the straight direction. The target direction is marked in green. The lane with the 3D bounding box of motor vehicles is set to 1, and the lane without the 3D bounding box of motor vehicles is set to 0. When a vehicle crosses two lanes, the lane with the larger position of the 3D bounding box of motor vehicles is the target lane of the vehicle. Count the number of vehicles in the left turn direction and the number of vehicles in the straight direction for each lane. Step S22: Create a 3D bounding box for non-motorized vehicles and pedestrians in the non-motorized vehicle lane. Predict the target direction of non-motorized vehicles and pedestrians based on the vehicle turning direction and traffic light status. Set a threshold to determine the target direction. When the current prediction accuracy is greater than the threshold, mark the target direction in green. When the prediction accuracy is less than the threshold, mark the target direction in yellow. Count the number of vehicles turning left and non-motorized vehicles and pedestrians going straight. Analyze the probability of turning left and going straight based on the road history data. Estimate the number of non-motorized vehicles and pedestrians marked with yellow target directions in the left and straight directions. Step S23: Based on the following conditions, when the number of 3D boxes in the left-turn direction is not 0, the traffic signal left-turn light for that lane is 1; when the number of 3D boxes in the left-turn direction is 0, the traffic signal left-turn light for that lane is 0; when the number of 3D boxes in the straight-ahead direction is not 0, the traffic signal straight-ahead light for that lane is 1; when the number of 3D boxes in the straight-ahead direction is 0, the traffic signal straight-ahead light for that lane is 0, construct the dynamic traffic flow map of the current road.

7. The real-time monitoring system based on cloud computing according to claim 1, characterized in that: Step S4 further includes the following steps: Step S41: Based on the traffic signal left turn light and the traffic signal straight light markings and the corresponding number of three-dimensional boxes, calculate the ratio between the number of three-dimensional boxes in the current prohibited direction and the number of traveling direction. When the ratio is greater than a threshold, increase the green light timing for the prohibited direction. When the ratio is less than 20% of the threshold, follow the original traffic light timing. When the ratio is less than 20% of the threshold, increase the green light timing for the traveling direction. Adjust the time period according to the different colors of the calendar markings. Step S42: Based on historical road videos and recorded adjustment time periods, extract the nodes and key features of the time periods, predict the traffic flow in each future time period, provide corresponding traffic light timings, and formulate intelligent traffic decision rules.

8. The cloud computing-based real-time monitoring system according to claim 1, characterized in that: Step S5 further includes: identifying the traffic light markings based on the dynamic traffic flow map; when the traffic light marking for the direction of travel is 0, ending the green light for the current direction of travel and switching to the next traffic light marking 1 in the original traffic signal rule switching sequence; when all traffic lights at the intersection are marked 0, following the original traffic signal rules.

Images