Tunnel dynamic illumination method based on computer vision

Abstract

The invention relates to a tunnel dynamic illumination method based on computer vision, and the method comprises the following steps: arranging nodes in a tunnel, carrying out the image collection of each node, and generating collection information; each node processes the collected information by using a vehicle detection algorithm, generates target detection information and transmits the target detection information to the server; the server processes the target detection information transmitted by each node, and calculates the instantaneous speed, the average speed and the driving direction of the vehicle according to the fixed distance between the nodes and the time difference of the vehicle between different nodes; the server constructs a real-time traffic flow information map according to the instantaneous speed, the average speed and the driving direction of the vehicle, predicts the vehicle speed according to the traffic flow information map, and generates control information; and the control center controls a plurality of lamps in the tunnel to turn on or turn off light according to the control information. Compared with the prior art, the vehicle detection accuracy can be improved, intelligent lighting is achieved, and energy consumption is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the field of lighting technology, and in particular to a dynamic tunnel lighting method based on computer vision. Background Art

[0002] In modern transportation systems, tunnels are crucial sections of the road, and the performance of their lighting systems is directly related to driving safety and energy efficiency. Currently, traditional tunnel lighting systems have numerous shortcomings in vehicle sensing technology. Existing vehicle sensing primarily relies on single-point triggering devices such as radar, coils, or infrared sensors. These devices can only determine whether a vehicle has passed through a specific area, but are unable to continuously and accurately track its position and motion. For example, when a vehicle is traveling in a tunnel, traditional systems struggle to obtain precise real-time location information, making it impossible to provide timely and accurate data support for lighting control.

[0003] Furthermore, existing lighting systems struggle to adapt to the dynamics of actual traffic flow. Unable to detect changes in traffic flow in a timely manner, lighting control often lags behind. During periods of low traffic, the lighting system may maintain high brightness, wasting energy. However, during periods of heavy traffic, insufficient lighting can compromise driving safety. This delayed lighting control not only results in high energy consumption but also compromises driving safety and increases the potential risk of traffic accidents. Summary of the Invention

[0004] The present invention aims to at least partially address one of the aforementioned technical problems in the prior art. To this end, one objective of the present invention is to propose a computer vision-based dynamic tunnel lighting method that improves vehicle detection accuracy, enables intelligent lighting, and significantly reduces energy consumption.

[0005] The present invention solves the above technical problems with the following technical solutions: A tunnel dynamic lighting method based on computer vision, comprising the following steps: Nodes are arranged at each lamp in the tunnel. Each node collects images of vehicles traveling in the tunnel and generates collected information. Each node processes the collected information using a vehicle detection algorithm, generates target detection information, and transmits it to the server. The server processes the target detection information transmitted by each node and calculates the vehicle's instantaneous speed, average speed, and direction based on the fixed distance between nodes and the time difference between the vehicle's appearance at different nodes. The server constructs a real-time traffic flow information map based on the vehicle's instantaneous speed, average speed, and direction, predicts vehicle speed based on the traffic flow information map, and generates control information. The control center controls multiple lamps in the tunnel to turn on or off the lights according to the control information.

[0006] The beneficial effects of the present invention are: the nodes, servers, control center and lamps work in coordination to detect vehicles in the tunnel, thereby improving the accuracy of vehicle detection; it can also reduce network load and improve data processing efficiency; it also facilitates system expansion; realizes intelligent lighting and significantly reduces energy consumption.

[0007] On the basis of the above technical solution, the present invention can also be improved as follows.

[0008] Furthermore, the node includes: A camera is used to capture images of vehicles traveling in the tunnel and generate captured information; A microcontroller is connected to the camera via a line, and the microcontroller is connected to the server using the communication module provided therein; the microcontroller is used to process the collected information using a vehicle detection algorithm, generate target detection information and transmit it to the server.

[0009] The beneficial effects of adopting the above further solution are: the camera and microcontroller improve the comprehensiveness and accuracy of vehicle detection, avoiding detection blind spots; ensuring that the node can work reliably in complex tunnel environments without interference from other equipment, thereby improving stability.

[0010] Furthermore, the target detection information adopts a structured data format, including a detection timestamp, a vehicle type, and a node ID.

[0011] The beneficial effect of adopting the above further solution is that structured data has the characteristics of small data volume and low upload frequency, which can not only ensure real-time performance but also ensure the stability of data transmission, greatly reducing the network transmission burden and improving data processing efficiency.

[0012] Furthermore, the server determines whether the vehicle has abnormal behavior based on the vehicle's instantaneous speed, average speed and driving direction, and generates abnormal information when it is determined that the vehicle is parked, reversed or driving at a low speed; at the same time, the traffic density in the area where the vehicle is located is counted in real time, statistical information is generated, and the abnormal information and statistical information are transmitted to the traffic management department.

[0013] The beneficial effect of adopting the above further solution is that the driving trajectories of all vehicles in the tunnel can be clearly tracked through the traffic information map, and the vehicle speed can be accurately predicted, providing an important basis for subsequent lighting control and traffic management.

[0014] Furthermore, the control center controls multiple lamps in the tunnel to turn on or off the lights according to the control information, specifically including the following steps: The server predicts the vehicle speed and direction based on the traffic flow information map and generates control information; the control center controls the lamps in the area where the vehicle is about to enter in the tunnel to turn on the lights based on the control information; The server predicts the vehicle speed and direction based on the traffic information map and generates control information; the control center controls the lamps in the tunnel in the set time period without vehicles entering the area to turn off the lights based on the control information.

[0015] The beneficial effect of adopting the above further solution is: when a vehicle is detected to be about to enter the target area, the lights are turned on in advance to provide the driver with sufficient vision and ensure the safe entry of the vehicle; if no subsequent vehicle is detected entering the target area within the set time, the lights are turned off, effectively saving energy. This lighting control based on the actual passing of the vehicle avoids unnecessary energy waste.

[0016] Furthermore, the control center controls the multiple lamps in the tunnel to turn on or off the lights according to the control information, and specifically includes the following steps: The server predicts the vehicle speed and direction based on the traffic information map, and the control center controls the lamps in the area where the traffic density in the tunnel exceeds the set threshold to keep on turning on the lights based on the control information.

[0017] The beneficial effect of adopting the above further solution is that the server controls the lamps in the target area through the control center to keep the lights on, ensuring the continuity of the driver's field of vision and ensuring that the driver will not be visually disturbed by lighting interruptions during driving.

[0018] Furthermore, the control center controls the multiple lamps in the tunnel to turn on or off the lights according to the control information, and specifically includes the following steps: The server predicts the vehicle speed and direction based on the traffic information map, and the control center controls the lamps in the tunnel where the vehicle is stagnant or reversing to flash and perform emergency lighting based on the control information.

[0019] The beneficial effect of adopting the above further solution is: by controlling the lamps to flash and turn on the emergency lighting, the driver is reminded to pay attention to safety, and at the same time, convenience is provided for subsequent rescue work. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a flow chart of a computer vision-based tunnel dynamic lighting method of the present invention; Figure 2 This is a module block diagram of the node and server of the present invention.

[0021] In the accompanying drawings, the components represented by the reference numerals are as follows: 1. Node, 2. Server, 3. Control center, 4. Lighting. DETAILED DESCRIPTION

[0022] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0023] like Figure 1 and Figure 2 As shown, a tunnel dynamic lighting method based on computer vision includes the following steps: Nodes 1 are arranged at corresponding locations of each lamp 4 in the tunnel. Each node 1 collects images of vehicles traveling in the tunnel and generates collected information. Each node 1 processes the collected information using a vehicle detection algorithm and generates target detection information that is transmitted to server 2. Server 2 processes the target detection information transmitted by each node 1 and calculates the vehicle's instantaneous speed, average speed, and direction of travel based on the fixed distance between nodes 1 and the time difference between the vehicle's appearance at different nodes 1. Server 2 constructs a real-time traffic flow information map based on the vehicle's instantaneous speed, average speed, and direction of travel, predicts vehicle speed based on the traffic flow information map, and generates control information. The control center 3 controls the multiple lamps 4 in the tunnel to turn on or off the lights according to the control information.

[0024] In the specific application of this embodiment, in order to enable the microcontroller in node 1 to efficiently run the vehicle detection algorithm, a lightweight vehicle detection algorithm, such as one based on MobileNetV1 or a custom convolutional network, is deployed on a microcontroller such as ESP32-S3. Before deployment, the vehicle detection algorithm is pruned and quantized to adapt to the limited computing power and memory constraints of the microcontroller by removing redundant parameters and optimizing data representation.

[0025] The algorithm is trained using frameworks such as ESP-DL. During the training process, a large amount of tunnel scene data is collected, including data from complex scenarios such as low illumination, strong light reflection, and exhaust gas interference. By enhancing and adapting this data, the algorithm model is continuously optimized to improve its recognition accuracy and robustness. When necessary, a knowledge distillation strategy is introduced to transfer the capabilities of the high-precision model to the micro-model, further improving the performance of the micro-model.

[0026] This embodiment achieves continuous perception of vehicle position and speed, breaking away from dependence on traditional radar and coil systems; through visual recognition technology, it can accurately obtain real-time information on the vehicle's position and motion status in the tunnel, providing reliable data support for lighting control and traffic management.

[0027] The microcontroller in node 1 greatly reduces the communication and computing load when completing target detection; it reduces the transmission of large amounts of raw image data and reduces network bandwidth requirements; at the same time, local computing reduces the computing pressure of the central server and improves the operating efficiency of the entire system.

[0028] The data reported by Node 1 is lightweight and structured, making it ideal for centralized computing. The computing server can quickly analyze and process this data, generating traffic flow information maps in a timely manner, providing strong support for dynamic lighting control and traffic management decisions. It is highly scalable and can adapt to the deployment requirements of tunnels of any length and structure. Whether it is a short tunnel or a long tunnel, a straight tunnel or a curved tunnel, the system can be quickly deployed and efficiently operated by simply adjusting the layout spacing and number of device nodes according to actual conditions.

[0029] Supports dynamic energy-saving lighting strategies, significantly improving the intelligence level of the lighting system and reducing energy consumption; automatically adjusts lighting brightness and on / off status according to vehicle driving conditions and traffic density, minimizing energy waste while ensuring driving safety.

[0030] This invention significantly reduces costs while ensuring perception accuracy, providing key support for smart tunnel lighting and traffic control, and has broad application prospects and significant economic benefits.

[0031] In the above embodiment, the node 1 includes: A camera is used to capture images of vehicles traveling in the tunnel and generate captured information; A microcontroller is connected to the camera via a line, and the microcontroller is connected to the server 2 using the communication module provided therein; the microcontroller is used to process the collected information using the vehicle detection algorithm, generate target detection information and transmit it to the server 2.

[0032] In the specific application of this embodiment, when deploying nodes 1 in the tunnel, full consideration is given to the comprehensiveness and accuracy of vehicle detection. Each node 1 consists of a microcontroller and a camera, and is installed near the lighting fixtures 4 in the tunnel or on a fixed bracket. The layout spacing of the nodes 1 is flexibly determined based on the detection accuracy requirements. For example, a group is deployed every 3 meters to ensure that every lane in the tunnel can be covered by the effective field of view of the camera. Such a layout design can capture vehicle information in all directions and avoid detection blind spots.

[0033] Each node 1 uses an independent power supply to ensure its stable operation; at the same time, node 1 has wireless or wired communication capabilities, which enables node 1 to complete image acquisition and recognition calculations, and exchange data with server 2 through a network interface. Independent power supply and diversified communication methods ensure that node 1 can work reliably in complex tunnel environments without being interfered with by other equipment, thereby improving stability.

[0034] In the above embodiment, the target detection information adopts a structured data format, including a detection timestamp, a vehicle type, and a node 1 ID.

[0035] When this embodiment is applied specifically, structured data has the characteristics of small data volume and low upload frequency, which can ensure both real-time performance and stability of data transmission; compared with traditional transmission of large amounts of original image data, it greatly reduces the network transmission burden and improves data processing efficiency.

[0036] In the above embodiment, the server 2 determines whether the vehicle has abnormal behavior based on the vehicle's instantaneous speed, average speed and driving direction, and generates abnormal information when it is determined that the vehicle is parked, reversed or driving at a low speed; at the same time, the traffic density in the area where the vehicle is located is counted in real time, statistical information is generated, and the abnormal information and statistical information are transmitted to the traffic management department.

[0037] In the specific application of this embodiment, server 2 uses this data to construct a real-time traffic information map. Through the traffic information map, the driving trajectories of all vehicles in the tunnel can be clearly tracked and the vehicle speeds can be accurately predicted. At the same time, server 2 also has an intelligent analysis function, which can determine whether there are abnormal behaviors of vehicles, such as parking, reversing, driving at low speeds, etc., and perform real-time statistics on the traffic density in specific areas. These analysis results provide an important basis for subsequent lighting control and traffic management.

[0038] In the above embodiment, the control center 3 controls the multiple lamps 4 in the tunnel to turn on or off the lights according to the control information, specifically including the following steps: The server 2 predicts the vehicle speed and direction based on the traffic flow information map and generates control information; the control center 3 controls the lamps 4 in the area where the vehicle is about to enter in the tunnel to turn on the lights according to the control information; The server 2 predicts the vehicle speed and direction based on the traffic information map and generates control information; the control center 3 controls the lamps 4 in the tunnel that are not in the vehicle entry area within the set time to turn off the lights based on the control information.

[0039] In a specific application of this embodiment, when it is detected that a vehicle is about to enter a target area, the server 2 controls the lamps 4 in the target area through the control center 3 to turn on the lights in advance, providing the driver with sufficient vision and ensuring the safe entry of the vehicle. For example, when the vehicle is 50 meters away from the target area, the control center 3 can issue a command to turn on the lights. After a vehicle passes through the target area, if no subsequent vehicle is detected entering the target area within the set time, the server 2 controls the lamps 4 in the target area through the control center 3 to turn off the lights, effectively saving energy. This lighting control based on the actual passage of vehicles avoids unnecessary energy waste.

[0040] In the above embodiment, the control center 3 controls multiple lamps 4 in the tunnel to turn on or off the lights according to the control information, and specifically includes the following steps: the server 2 predicts the vehicle speed and direction according to the traffic information map, and the control center 3 controls the lamps 4 in the area of ​​the tunnel where the traffic density exceeds the set threshold to keep on turning on the lights according to the control information.

[0041] In the specific application of this embodiment, when there is dense traffic in the target area, in order to ensure the continuity of the driver's field of vision, the server 2 controls the lamps 4 in the target area through the control center 3 to keep the lights on; for example, when the traffic density in the target area reaches the set threshold and more than 20 vehicles pass through per minute, the lights are kept on to ensure that the driver will not be visually disturbed by lighting interruptions during driving.

[0042] In the above embodiment, the control center 3 controls the multiple lamps 4 in the tunnel to turn on or off the lights according to the control information, and specifically includes the following steps: The server 2 predicts the vehicle speed and direction based on the traffic information map, and the control center 3 controls the lamps 4 in the tunnel where the vehicle is stagnant or reversing to flash and perform emergency lighting according to the control information.

[0043] When this embodiment is specifically applied, for abnormal events, such as when a vehicle is stagnant or reverses in the target area, the control center 3 controls the lamps 4 in the tunnel where the vehicle is stagnant or reverses to flash according to the control information, and turns on the emergency lighting to remind the driver to pay attention to safety, while providing convenience for subsequent rescue work.

[0044] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A tunnel dynamic lighting method based on computer vision, characterized in that: The following steps are involved: Nodes are arranged at each lamp in the tunnel. Each node collects images of vehicles traveling in the tunnel and generates collected information. Each node processes the collected information using a vehicle detection algorithm, generates target detection information, and transmits it to the server. The server processes the target detection information transmitted by each node and calculates the vehicle's instantaneous speed, average speed, and direction based on the fixed distance between nodes and the time difference between the vehicle's appearance at different nodes. The server constructs a real-time traffic flow information map based on the vehicle's instantaneous speed, average speed, and direction, predicts vehicle speed based on the traffic flow information map, and generates control information. The control center controls multiple lamps in the tunnel to turn on or off the lights according to the control information.

2. The computer vision-based tunnel dynamic lighting method according to claim 1, characterized in that: The nodes include: A camera is used to capture images of vehicles traveling in the tunnel and generate captured information; A microcontroller is connected to the camera via a line, and the microcontroller is connected to the server using the communication module provided therein; the microcontroller is used to process the collected information using a vehicle detection algorithm, generate target detection information and transmit it to the server.

3. The computer vision-based tunnel dynamic lighting method according to claim 1, characterized in that: The target detection information adopts a structured data format, including a detection timestamp, a vehicle type, and a node ID.

4. The computer vision-based tunnel dynamic lighting method according to claim 1, characterized in that: The server determines whether the vehicle has abnormal behavior based on the vehicle's instantaneous speed, average speed and driving direction. When it is determined that the vehicle is parked, reversing or driving at a low speed, it generates abnormal information; at the same time, it conducts real-time statistics on the traffic density in the area where the vehicle is located, generates statistical information, and transmits the abnormal information and statistical information to the traffic management department.

5. The computer vision-based tunnel dynamic lighting method according to claim 1, characterized in that: The control center controls multiple lamps in the tunnel to turn on or off the lights according to the control information, which specifically includes the following steps: The server predicts the vehicle speed and direction based on the traffic flow information map and generates control information; the control center controls the lamps in the area where the vehicle is about to enter in the tunnel to turn on the lights based on the control information; The server predicts the vehicle speed and direction based on the traffic information map and generates control information; the control center controls the lamps in the tunnel in the set time period without vehicles entering the area to turn off the lights based on the control information.

6. The computer vision-based tunnel dynamic lighting method according to claim 5, characterized in that: The control center controls multiple lamps in the tunnel to turn on or off the lights according to the control information, and specifically includes the following steps: The server predicts the vehicle speed and direction based on the traffic information map, and the control center controls the lamps in the area where the traffic density in the tunnel exceeds the set threshold to keep on turning on the lights based on the control information.

7. The computer vision-based tunnel dynamic lighting method according to claim 6, characterized in that: The control center controls multiple lamps in the tunnel to turn on or off the lights according to the control information, and specifically includes the following steps: The server predicts the vehicle speed and direction based on the traffic information map, and the control center controls the lamps in the tunnel where the vehicle is stagnant or reversing to flash and perform emergency lighting based on the control information.

Images