A holographic intersection signal real-time optimization method and device based on edge computing

Abstract

The present application relates to traffic signal optimization, in particular to a holographic intersection signal real-time optimization method and device based on edge computing, video stream data and radar data of each intersection are obtained, and the video stream data and radar data are processed; the edge base station of each intersection calculates the corresponding intersection traffic volume based on the corresponding video stream data and radar data; the edge base station of each intersection dynamically learns and adjusts according to the intersection traffic volume, obtains the first traffic signal timing strategy, and uploads to the cloud server; the cloud server globally optimizes the first traffic signal timing strategy uploaded by each edge base station according to the intersection traffic volume information of each intersection, and obtains the second traffic signal timing strategy; the cloud server sends the second traffic signal timing strategy to the signal machine of each intersection, and each signal machine executes the second traffic signal timing strategy; the technical scheme provided by the present application can effectively overcome the defect that the traffic signal cannot be reasonably and real-time optimized.

Description

Technical Field

[0001] This invention relates to traffic signal optimization, specifically to a method and apparatus for real-time optimization of holographic intersection signals based on edge computing. Background Technology

[0002] Holographic intersections utilize both intersection radar and checkpoint cameras for perception. While maintaining the original non-on-site enforcement functions, they integrate the latest sensor technology, high-precision map technology, AI algorithms, high-performance chips, and edge computing technology to build "intelligence + perception" capabilities. This generates a variety of accurate, efficient, and real-time metadata, such as vehicle spatiotemporal data, vehicle identity, violation capture, decimeter-level vehicle trajectories, and traffic light status. This lays a complete data foundation for refined intersection management and reduces the computational, storage, spatial, and network transmission pressure on the central side.

[0003] With the increasing number of vehicles on the road, the traffic burden on urban roads has grown, and the inability of traffic lights to effectively regulate traffic flow in all directions has led to increasingly severe traffic congestion. Optimizing traffic lights is therefore crucial for improving road traffic efficiency.

[0004] Currently, some mathematical models, such as the Webster model, are used for traffic light timing strategy design. Although these models are simple and easy to use, they have poor adaptability and are difficult to apply to intersections in different environments. With the development of artificial neural network algorithms, traffic light timing algorithms based on them have emerged one after another, such as the traffic light timing algorithm based on BP neural network. Although it has very strong adaptive and learning capabilities, it is difficult to apply to the real-time optimization control of traffic lights because it requires a large amount of data for learning. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] In view of the above-mentioned shortcomings of the existing technology, the present invention provides a method and device for real-time optimization of holographic intersection signals based on edge computing, which can effectively overcome the defect of the existing technology that cannot perform reasonable real-time optimization of traffic lights.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] A real-time optimization method for holographic intersection signals based on edge computing includes the following steps:

[0010] S1. Acquire video stream data and radar data from each intersection, and perform data processing on the video stream data and radar data;

[0011] S2. The edge base stations at each intersection calculate the corresponding traffic flow based on the corresponding video stream data and radar data.

[0012] S3. The edge base stations at each intersection dynamically learn and adjust according to the traffic flow at the intersection to obtain the first traffic light timing strategy and upload it to the cloud server.

[0013] S4. The cloud server performs global optimization of the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information of each intersection, and obtains the second traffic light timing strategy.

[0014] S5, the cloud server sends the second traffic light timing strategy to the traffic signal controllers at each intersection, and each traffic signal controller executes the second traffic light timing strategy.

[0015] Preferably, the edge base stations at each intersection in S2 calculate the corresponding traffic flow at the intersection based on the corresponding video stream data and radar data, including:

[0016] The edge base station calculates and tracks the traffic flow at the first intersection based on the video stream data of the intersection, and calculates and tracks the traffic flow at the second intersection based on the radar data of the intersection.

[0017] The edge base station combines the traffic flow at the first intersection with the traffic flow at the second intersection to obtain the real-time traffic flow at the intersection.

[0018] Preferably, the edge base station calculates and tracks the traffic flow at the first intersection based on the video stream data of the intersection, including:

[0019] Whether to perform background updates is determined based on the ratio between inter-frame differences and background differences;

[0020] Extract the foreground image and remove shadows using the HSV feature method;

[0021] The image was transformed using Hough transform, and the minimum bounding rectangle method was used to calculate and track each target vehicle to obtain the traffic flow at the first intersection.

[0022] Preferably, the edge base station calculates and tracks the traffic flow at the second intersection based on radar data from the intersection, including:

[0023] The intermediate frequency signal is tracked and processed to obtain the distance information of each target vehicle;

[0024] Plot the trajectory of each target vehicle with time as the horizontal axis and distance as the vertical axis, and then convert the trajectory diagram into a binary image.

[0025] The binary image is divided into several sub-images according to time intervals, and each sub-image is divided into several sub-regions according to distance. The traffic flow at the second intersection is calculated and tracked based on the black pixel values ​​in each sub-region.

[0026] Preferably, the step of dividing the binary image into several sub-images according to time intervals, dividing each sub-image into several sub-regions according to distance, and calculating and tracking the traffic flow at the second intersection based on the black pixel values ​​in each sub-region includes:

[0027] After dividing each sub-image into several sub-regions according to distance, only the target region containing the distance segment [p1, p2] in each sub-image is retained, and the sum of the black pixel values ​​S in the target region is calculated. i ;

[0028] The traffic flow f2 at the second intersection is calculated using the following formula based on the sum of the black pixel values ​​in the target region of each sub-image:

[0029]

[0030] Where p1 is the starting position of the radar detection area and p2 is the ending position of the radar detection area;

[0031] S i th = T / t, where T is the time interval corresponding to dividing the binary image into several sub-images, t is the radar data processing cycle, and floor(·) represents rounding down.

[0032] Preferably, the edge base stations at each intersection in S3 dynamically learn and adjust based on the traffic flow at the intersection, including:

[0033] The dynamic learning adjustment formula is as follows:

[0034] T t+1 (s t+1 ,d t+1 )=(F-λ)·T t (s t ,d t )+λ·[r t+1 +β·maxT t+1 (s t+1 ,d t+1 )]

[0035] Where s represents the traffic state at the intersection, d represents the traffic signal timing strategy, T is a table of s and d, F represents the real-time traffic flow at the intersection, λ represents the learning efficiency, β represents the discount factor, and r represents the feedback after executing the traffic signal timing strategy d.

[0036] Preferably, in S4, the cloud server globally optimizes the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, resulting in a second traffic light timing strategy, including:

[0037] The cloud server uses a preset algorithm to globally optimize the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, and obtains the second traffic light timing strategy.

[0038] The preset algorithm is one of the following: genetic algorithm, ant colony algorithm, simulated annealing algorithm, and particle swarm algorithm.

[0039] A real-time optimization device for holographic intersection signals based on edge computing includes a video stream data acquisition device, a radar device, a data processing module, an edge base station, and a cloud server;

[0040] Video stream data acquisition devices are installed at each intersection to collect video stream data from each intersection.

[0041] Radar devices are installed at various intersections to collect radar data from each intersection.

[0042] The data processing module is used to process the video stream data and radar data from each intersection.

[0043] Edge base stations are set up at each intersection. They calculate the traffic flow at the corresponding intersection based on the corresponding video stream data and radar data, and dynamically learn and adjust according to the traffic flow to obtain the first traffic light timing strategy, which is then uploaded to the cloud server.

[0044] The cloud server performs global optimization of the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, obtains the second traffic light timing strategy, and sends the second traffic light timing strategy to the traffic signal controllers at each intersection.

[0045] Preferably, the radar device includes a radio frequency module, a receiving antenna, and a mixer;

[0046] The radio frequency module transmits radio frequency signals to the road.

[0047] The receiving antenna receives the radio frequency signals reflected by each target vehicle.

[0048] The mixer down-converts the reflected signals from each target vehicle to obtain an intermediate frequency signal.

[0049] (III) Beneficial Effects

[0050] Compared with existing technologies, the real-time optimization method and apparatus for holographic intersection signals based on edge computing provided by this invention has the following advantages:

[0051] 1) Acquire video stream data and radar data from each intersection, and process the video stream data and radar data. The edge base stations at each intersection calculate the corresponding traffic flow based on the corresponding video stream data and radar data. The edge base stations at each intersection dynamically learn and adjust according to the traffic flow to obtain the first traffic light timing strategy and upload it to the cloud server. The edge base stations can accurately calculate the traffic flow at the intersection based on the video stream data and radar data, and dynamically learn and adjust according to the traffic flow to obtain the adaptive first traffic light timing strategy for each intersection.

[0052] 2) The cloud server performs global optimization of the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information of each intersection, and obtains the second traffic light timing strategy. The cloud server sends the second traffic light timing strategy to the traffic signal controllers at each intersection, and each traffic signal controller executes the second traffic light timing strategy. The cloud server can perform global optimization of the uploaded first traffic light timing strategy based on the traffic flow information of each intersection, and obtain the second traffic light timing strategy adapted to the traffic flow situation of each intersection, thereby realizing reasonable real-time optimization of the holographic intersection traffic lights. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0054] Figure 1 This is a schematic diagram of the process of the present invention;

[0055] Figure 2 This is a schematic diagram illustrating the process by which an edge base station calculates the corresponding traffic flow at an intersection based on corresponding video stream data and radar data in this invention. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0057] A real-time optimization method for holographic intersection signals based on edge computing, such as... Figure 1 and Figure 2As shown, ① acquire video stream data and radar data from each intersection, and perform data processing on the video stream data and radar data.

[0058] ② The edge base stations at each intersection calculate the corresponding traffic flow based on the corresponding video stream data and radar data, specifically including:

[0059] The edge base station calculates and tracks the traffic flow at the first intersection based on the video stream data of the intersection, and calculates and tracks the traffic flow at the second intersection based on the radar data of the intersection.

[0060] The edge base station combines the traffic flow at the first intersection with the traffic flow at the second intersection to obtain the real-time traffic flow at the intersection.

[0061] 1) The edge base station calculates and tracks the traffic flow at the first intersection based on the video stream data of the intersection, including:

[0062] Whether to perform background updates is determined based on the ratio between inter-frame differences and background differences;

[0063] Extract the foreground image and remove shadows using the HSV feature method;

[0064] The image was transformed using Hough transform, and the minimum bounding rectangle method was used to calculate and track each target vehicle to obtain the traffic flow at the first intersection.

[0065] 2) The edge base station calculates and tracks the traffic flow at the second intersection based on radar data from the intersection, including:

[0066] The intermediate frequency signal is tracked and processed to obtain the distance information of each target vehicle;

[0067] Plot the trajectory of each target vehicle with time as the horizontal axis and distance as the vertical axis, and then convert the trajectory diagram into a binary image.

[0068] The binary image is divided into several sub-images according to time intervals, and each sub-image is divided into several sub-regions according to distance. The traffic flow at the second intersection is calculated and tracked based on the black pixel values ​​in each sub-region.

[0069] Specifically, the binary image is divided into several sub-images according to time intervals, and each sub-image is further divided into several sub-regions according to distance. The traffic flow at the second intersection is then calculated and tracked based on the black pixel values ​​in each sub-region, including:

[0070] After dividing each sub-image into several sub-regions according to distance, only the target region containing the distance segment [p1, p2] in each sub-image is retained, and the sum of the black pixel values ​​S in the target region is calculated. i ;

[0071] The traffic flow f2 at the second intersection is calculated using the following formula based on the sum of the black pixel values ​​in the target region of each sub-image:

[0072]

[0073] Where p1 is the starting position of the radar detection area and p2 is the ending position of the radar detection area;

[0074] S i th = T / t, where T is the time interval corresponding to dividing the binary image into several sub-images, t is the radar data processing cycle, and floor(·) represents rounding down.

[0075] ③ The edge base stations at each intersection dynamically learn and adjust based on the traffic flow at the intersection to obtain the first traffic light timing strategy and upload it to the cloud server.

[0076] Specifically, the edge base stations at each intersection dynamically learn and adjust based on traffic flow at the intersection, including:

[0077] The dynamic learning adjustment formula is as follows:

[0078] T t+1 (s t+1 ,d t+1 )=(F-λ)·T t (s t ,d t )+λ·[r t+1 +β·maxT t+1 (s t+1 ,d t+1 )]

[0079] Where s represents the traffic state at the intersection, d represents the traffic signal timing strategy, T is a table of s and d, F represents the real-time traffic flow at the intersection, λ represents the learning efficiency, β represents the discount factor, and r represents the feedback after executing the traffic signal timing strategy d.

[0080] The above technical solution acquires video stream data and radar data from each intersection, processes the video stream data and radar data, and the edge base stations at each intersection calculate the corresponding traffic flow based on the corresponding video stream data and radar data. The edge base stations at each intersection dynamically learn and adjust according to the traffic flow to obtain the first traffic light timing strategy, and upload it to the cloud server. The edge base stations can accurately calculate the traffic flow at the intersection based on the video stream data and radar data, and dynamically learn and adjust according to the traffic flow to obtain the adaptive first traffic light timing strategy for each intersection.

[0081] like Figure 1As shown, ④ the cloud server performs global optimization of the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, resulting in the second traffic light timing strategy, which specifically includes:

[0082] The cloud server uses a preset algorithm to globally optimize the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, and obtains the second traffic light timing strategy.

[0083] The preset algorithm is one of the following: genetic algorithm, ant colony algorithm, simulated annealing algorithm, and particle swarm algorithm.

[0084] ⑤ The cloud server sends the second traffic light timing strategy to the traffic signal controllers at each intersection, and each traffic signal controller executes the second traffic light timing strategy.

[0085] The above technical solution involves the cloud server globally optimizing the first traffic signal timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, resulting in a second traffic signal timing strategy. The cloud server then sends the second traffic signal timing strategy to the traffic signal controllers at each intersection, and each traffic signal controller executes the second traffic signal timing strategy. The cloud server can globally optimize the uploaded first traffic signal timing strategy based on the traffic flow information at each intersection, resulting in a second traffic signal timing strategy that adapts to the traffic flow conditions at each intersection, thereby achieving reasonable real-time optimization of the holographic intersection traffic signals.

[0086] This application also discloses a real-time optimization device for holographic intersection signals based on edge computing, including a video stream data acquisition device, a radar device, a data processing module, an edge base station, and a cloud server;

[0087] Video stream data acquisition devices are installed at each intersection to collect video stream data from each intersection.

[0088] Radar devices are installed at various intersections to collect radar data from each intersection.

[0089] The data processing module is used to process the video stream data and radar data from each intersection.

[0090] Edge base stations are set up at each intersection. They calculate the traffic flow at the corresponding intersection based on the corresponding video stream data and radar data, and dynamically learn and adjust according to the traffic flow to obtain the first traffic light timing strategy, which is then uploaded to the cloud server.

[0091] The cloud server performs global optimization of the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, obtains the second traffic light timing strategy, and sends the second traffic light timing strategy to the traffic signal controllers at each intersection.

[0092] Specifically, the radar device includes a radio frequency module, a receiving antenna, and a mixer;

[0093] The radio frequency module transmits radio frequency signals to the road.

[0094] The receiving antenna receives the radio frequency signals reflected by each target vehicle.

[0095] The mixer down-converts the reflected signals from each target vehicle to obtain an intermediate frequency signal.

[0096] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A real-time optimization method for holographic intersection signals based on edge computing, characterized in that: Includes the following steps: S1. Acquire video stream data and radar data from each intersection, and perform data processing on the video stream data and radar data; S2. The edge base stations at each intersection calculate the corresponding traffic flow based on the corresponding video stream data and radar data, including: The edge base station calculates and tracks the traffic flow at the first intersection based on the video stream data of the intersection, and calculates and tracks the traffic flow at the second intersection based on the radar data of the intersection. The edge base station combines the traffic flow at the first intersection with the traffic flow at the second intersection to obtain the real-time traffic flow at the intersection. S3. The edge base stations at each intersection dynamically learn and adjust according to the traffic flow at the intersection to obtain the first traffic light timing strategy and upload it to the cloud server. S4. The cloud server performs global optimization of the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information of each intersection, and obtains the second traffic light timing strategy. S5. The cloud server sends the second traffic light timing strategy to the traffic signal controllers at each intersection, and each traffic signal controller executes the second traffic light timing strategy. In S2, the edge base stations calculate and track traffic flow at the second intersection based on radar data from the intersection, including: The intermediate frequency signal is tracked and processed to obtain the distance information of each target vehicle; Plot the trajectory of each target vehicle with time as the horizontal axis and distance as the vertical axis, and then convert the trajectory diagram into a binary image. The binary image is divided into several sub-images according to time intervals, and each sub-image is further divided into several sub-regions according to distance. The traffic flow at the second intersection is then calculated and tracked based on the black pixel values ​​in each sub-region. Specifically, this includes: After dividing each sub-image into several sub-regions according to distance, only the target region containing the distance segment [p1, p2] in each sub-image is retained, and the sum of the black pixel values ​​S in the target region is calculated. i ; The traffic flow f2 at the second intersection is calculated using the following formula based on the sum of the black pixel values ​​in the target region of each sub-image: ； Where p1 is the starting position of the radar detection area and p2 is the ending position of the radar detection area; S i The sum of the black pixel values ​​in the target region of the i-th sub-image, i=1,2,3,…,N. T represents the time interval corresponding to dividing the binary image into several sub-images, and t represents the radar data processing period. Indicates rounding down; In S3, the edge base stations at each intersection dynamically learn and adjust based on traffic flow at the intersection, including: The dynamic learning adjustment formula is as follows: ； Where s represents the traffic status at the intersection, d represents the traffic light timing strategy, T is a table of s and d, and F represents the real-time traffic flow at the intersection. For learning efficiency, denoted as the discount factor, and r as the feedback after implementing the traffic signal timing strategy d.

2. The real-time optimization method for holographic intersection signals based on edge computing according to claim 1, characterized in that: The edge base station calculates and tracks the traffic flow at the first intersection based on the video stream data of the intersection, including: Whether to perform background updates is determined based on the ratio between inter-frame differences and background differences; Extract the foreground image and remove shadows using the HSV feature method; The image was transformed using Hough transform, and the minimum bounding rectangle method was used to calculate and track each target vehicle to obtain the traffic flow at the first intersection.

3. The real-time optimization method for holographic intersection signals based on edge computing according to claim 1, characterized in that: In S4, the cloud server globally optimizes the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, resulting in a second traffic light timing strategy, including: The cloud server uses a preset algorithm to globally optimize the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, and obtains the second traffic light timing strategy. The preset algorithm is one of the following: genetic algorithm, ant colony algorithm, simulated annealing algorithm, and particle swarm algorithm.

4. A real-time optimization device for holographic intersection signals based on edge computing as described in claim 1, characterized in that: This includes video stream data acquisition devices, radar devices, data processing modules, edge base stations, and cloud servers; Video stream data acquisition devices are installed at each intersection to collect video stream data from each intersection. Radar devices are installed at various intersections to collect radar data from each intersection. The data processing module is used to process the video stream data and radar data from each intersection. Edge base stations are set up at each intersection. They calculate the traffic flow at the corresponding intersection based on the corresponding video stream data and radar data, and dynamically learn and adjust according to the traffic flow to obtain the first traffic light timing strategy, which is then uploaded to the cloud server. The cloud server performs global optimization of the first traffic light timing strategy uploaded by each edge base station based on the traffic flow information at each intersection, obtains the second traffic light timing strategy, and sends the second traffic light timing strategy to the traffic signal controllers at each intersection.

5. The real-time optimization device for holographic intersection signals based on edge computing according to claim 4, characterized in that: The radar device includes a radio frequency module, a receiving antenna, and a mixer; The radio frequency module transmits radio frequency signals to the road. The receiving antenna receives the radio frequency signals reflected by each target vehicle. The mixer down-converts the reflected signals from each target vehicle to obtain an intermediate frequency signal.

Images