Municipal road street lamp control method and system
Through the combination of recurrent neural network and diffusion model, the monitoring video data around the street lights is used to dynamically adjust the brightness of the street lights, solving the problem of inaccurate brightness in traditional street light control methods, and improving safety and energy utilization efficiency.
Patent Information
- Application Number
- CN202411869865.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-04-09
- Filing Date
- 2024-12-18
- Publication Date
- 2025-01-14
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional street light control methods lack the ability to accurately perceive and analyze real-time changes in the environment, resulting in the brightness of street lights being too dark or too bright, affecting safety and energy utilization efficiency.
The recurrent neural network model is used to obtain the monitoring video around the street lights, determine the environmental image, lighting street light location, traffic flow and people flow information, and calculate the target lighting image and the light brightness of each street light through the diffusion model and the convolutional neural network to achieve dynamic adjustment.
Accurately determining the brightness of street lights improves the safety of pedestrians and vehicles, reduces energy waste and light pollution, and improves the comfort and efficiency of urban lighting.
Smart Images
Figure CN119317004A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of street lamps, and in particular to a control method and system for street lamps on municipal roads. Background Art
[0002] With the acceleration of urbanization, the intelligent management of urban lighting systems has become increasingly important. As the main facility for urban lighting, the control method of street lamps is directly related to the efficiency of energy utilization and the quality of life of urban residents. Traditional street lamp control methods usually rely on preset schedules or simple photosensitive sensors to switch street lamps on and off. In traditional street lamp control methods, due to the lack of accurate perception and analysis capabilities of real-time changes in the environment, the problem of street lamp brightness being too dim or too bright often occurs. When the brightness of street lamps is too dim, it will affect the safe passage of pedestrians and vehicles and increase the risk of traffic accidents; when the brightness of street lamps is too bright, it will lead to energy waste and light pollution problems, affecting the rest and quality of life of residents.
[0003] Therefore, how to accurately determine the brightness of street lamps is a problem that needs to be solved at present. Summary of the invention
[0004] The main technical problem solved by the present invention is how to accurately determine the brightness of a street lamp.
[0005] According to a first aspect, the present invention provides a method for controlling street lamps on municipal roads, comprising: acquiring surveillance videos around the street lamps; determining a current environment image, multiple locations of lighting street lamps in the current environment image, the number of vehicles, the vehicle speed, the number of pedestrians, and the pedestrian speed using an environmental information processing model based on the surveillance videos around the street lamps; determining a lighting demand based on the number of vehicles, the number of pedestrians, the vehicle speed, and the pedestrian speed; determining a target lighting image of the current environment using a diffusion model based on the current environment image and the lighting demand; determining the light brightness of each lighting street lamp in the current environment image using a light brightness determination model based on the current environment image, the target lighting image of the current environment, and multiple locations of lighting street lamps in the current environment image; and controlling each lighting street lamp in the current environment image based on the light brightness of each lighting street lamp in the current environment image.
[0006] Furthermore, determining the lighting demand based on the number of vehicles, the number of pedestrians, the vehicle speed, and the pedestrian speed includes: performing weighted summation of the number of vehicles, the number of pedestrians, the vehicle speed, and the pedestrian speed according to a preset weight coefficient to obtain the lighting demand.
[0007] Furthermore, the controlling of each street lamp in the current environment image based on the light brightness of each street lamp in the current environment image includes: sending the light brightness of each street lamp in the current environment image to a street lamp control system, and the street lamp control system sends a control instruction to control each street lamp in the current environment image.
[0008] Furthermore, the input of the diffusion model is the current environment image and the lighting requirement, and the output of the diffusion model is a target lighting image of the current environment.
[0009] Furthermore, the environmental information processing model is a recurrent neural network model.
[0010] According to a second aspect, the present invention provides a control system for a street lamp on a municipal road, comprising: An acquisition module is used to acquire surveillance videos around street lamps; An environmental information processing module, used to determine a current environmental image, positions of multiple lighting street lamps in the current environmental image, the number of vehicles, the speed of vehicles, the number of pedestrians, and the speed of pedestrians based on the surveillance video around the street lamp using an environmental information processing model; A lighting demand determination module, configured to determine the lighting demand based on the number of vehicles, the number of people, the speed of the vehicles, and the speed of the people; A target lighting image generation module, configured to determine a target lighting image of the current environment using a diffusion model based on the current environment image and the lighting requirement; A light brightness determination module, configured to determine the light brightness of each streetlight in the current environment image using a light brightness determination model based on the current environment image, the target lighting image of the current environment, and the positions of multiple streetlights in the current environment image; A control module is used to control each streetlight in the current environment image based on the light brightness of each streetlight in the current environment image.
[0011] Furthermore, the lighting demand determination module is also used to: perform weighted summation of the number of vehicles, the number of pedestrians, the speed of vehicles, and the speed of pedestrians according to a preset weight coefficient to obtain the lighting demand.
[0012] Furthermore, the control module is also used to: send the light brightness of each lighting street lamp in the current environment image to the street lamp control system, and the street lamp control system sends a control instruction to control each lighting street lamp in the current environment image.
[0013] Furthermore, the input of the diffusion model is the current environment image and the lighting requirement, and the output of the diffusion model is a target lighting image of the current environment.
[0014] Furthermore, the environmental information processing model is a recurrent neural network model.
[0015] The present invention provides a method and system for controlling street lamps on municipal roads. The method includes acquiring surveillance videos around the street lamps; determining a current environment image, multiple locations of lighting street lamps in the current environment image, the number of vehicles, the vehicle speed, the number of pedestrians, and the pedestrian speed using an environmental information processing model based on the surveillance videos around the street lamps; determining a lighting demand based on the number of vehicles, the number of pedestrians, the vehicle speed, and the pedestrian speed; determining a target lighting image of the current environment using a diffusion model based on the current environment image and the lighting demand; determining the light brightness of each lighting street lamp in the current environment image using a light brightness determination model based on the current environment image, the target lighting image of the current environment, and multiple locations of lighting street lamps in the current environment image; and controlling each lighting street lamp in the current environment image based on the light brightness of each lighting street lamp in the current environment image. The method can accurately determine the street lamp brightness. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 A schematic diagram of a flow chart of a method for controlling streetlights on a municipal road provided by an embodiment of the present invention; Figure 2 A schematic diagram of a control system for a street light on a municipal road provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0017] In an embodiment of the present invention, there is provided Figure 1 A method for controlling a street lamp on a municipal road is shown, and the method for controlling a street lamp comprises steps S1 to S6: Step S1, obtaining surveillance video around the street lamp.
[0018] Surveillance videos around street lamps refer to real-time video streams obtained through cameras or monitoring equipment installed around street lamps, which are used to monitor and record the surrounding environment.
[0019] Step S2, based on the surveillance video around the street lamp, use the environmental information processing model to determine the current environmental image, the positions of multiple lighting street lamps in the current environmental image, the number of vehicles, the speed of vehicles, the number of people, and the speed of people.
[0020] The environmental information processing model is a recurrent neural network model, the input of which is the surveillance video around the street lamp, and the output of which is the current environmental image, the positions of multiple lighting street lamps in the current environmental image, the number of vehicles, the speed of vehicles, the number of people, and the speed of people. The recurrent neural network model includes a recurrent neural network (RNN). The recurrent neural network model can process sequence data, capture the information of the sequence, and output the result based on the correlation relationship between the previous and next data in the sequence. The recurrent neural network includes a hidden layer and an output layer. Each neuron in the hidden layer receives the input and the hidden state of the previous moment as input at each time step, and generates the output of the current moment and the hidden state of the next moment. By processing the surveillance video around the street lamp in a continuous time period with the recurrent neural network model, it is possible to output the features that comprehensively consider the correlation relationship between the sequences at each time point, so that the output features are more accurate and comprehensive.
[0021] The current environment image refers to a static image of the surrounding environment at the current moment captured in the surveillance video.
[0022] The lighting street lamp position indicates the specific position coordinates of each lighting street lamp displayed in the monitoring video.
[0023] The number of vehicles refers to the number of vehicles counted after the surveillance video is processed by the environmental information processing model.
[0024] Traffic speed refers to the average speed of traffic calculated from the vehicle motion trajectory in the surveillance video through the environmental information processing model.
[0025] The number of people flow refers to the number of people flow obtained by processing and counting the surveillance video through the environmental information processing model.
[0026] The crowd flow speed refers to the average crowd flow speed calculated by processing the surveillance video through the environmental information processing model.
[0027] Surveillance video can be viewed as a time series data, where each frame contains information about the current environment. Recurrent neural networks can process video data frame by frame in chronological order, thereby capturing the temporal correlation and dynamic changes in the video sequence. Recurrent neural networks can memorize previously processed information and combine it with the information of the current frame for inference. When processing surveillance videos, recurrent neural networks can use the information of previous frames to understand the content of the current frame, thereby better understanding the traffic and pedestrian flow. Through the hidden state of the recurrent neural network, abstract features in the video sequence can be learned, which helps to extract key information such as vehicles and pedestrians. These features can help accurately identify various objects in the current environment image, including the location of lighting street lamps, the number of vehicles, the speed of vehicles, the number of pedestrians, the speed of pedestrians, etc. Recurrent neural networks can model the dynamic changes and trends in time series data, and model and predict the dynamic conditions of vehicle and pedestrian flow in surveillance videos. By analyzing video frames at different time points, recurrent neural networks can infer the real-time state of the environment, including vehicle and pedestrian flow conditions.
[0028] In one embodiment, the environmental information processing model includes a region segmentation layer, a vehicle flow information determination layer, a human flow information determination layer, an environmental image output layer, and a street lamp positioning layer. The region segmentation layer, the vehicle flow information determination layer, the human flow information determination layer, the environmental image output layer, and the street lamp positioning layer all include a recurrent neural network. The input of the region segmentation layer is the monitoring video around the street lamp, the output of the region segmentation layer is the monitoring video of the sidewalk area and the monitoring video of the roadway area, the input of the vehicle flow information determination layer is the monitoring video of the roadway area, the output of the vehicle flow information determination layer is the number of vehicles and the speed of vehicles, the input of the human flow information determination layer is the monitoring video of the sidewalk area, the output of the human flow information determination layer is the number of people and the speed of people, the input of the environmental image output layer is the monitoring video around the street lamp, the output of the environmental image output layer is the current environmental image, the input of the street lamp positioning layer is the current environmental image, and the output of the street lamp positioning layer is the positions of multiple lighting street lamps in the current environmental image. Using multiple layers can achieve parallel processing, each layer can process different information at the same time, can effectively extract and separate specific information, thereby accelerating the calculation process of the entire system. For example, the area segmentation layer can extract monitoring videos of the sidewalk area and the roadway area from the monitoring video, and the vehicle flow information determination layer and the pedestrian flow information determination layer can extract information such as the number of vehicles, vehicle speed, number of pedestrians and pedestrian speed from their respective monitoring videos.
[0029] Step S3, determining the lighting demand based on the number of vehicles, the number of people, the speed of the vehicles, and the speed of the people.
[0030] In some embodiments, the light brightness of each street lamp in the current environment image may be sent to a street lamp control system, and the street lamp control system sends a control instruction to control each street lamp in the current environment image.
[0031] The number and speed of traffic flow can reflect the vehicle density and vehicle running speed on the road. If the traffic volume is large and the speed is fast, brighter lighting may be required to ensure the driver's clear vision and improve road safety.
[0032] The number and speed of pedestrian flow can reflect the density and movement speed of pedestrian activities. When the flow of people is large and the speed is fast, better lighting may be needed to ensure the safety and convenience of pedestrians.
[0033] The lighting demand indicates the degree of demand for lighting brightness determined according to the traffic and pedestrian flow. The greater the lighting demand, the greater the lighting brightness required to meet the needs of traffic and pedestrian flow.
[0034] Step S4: determining a target lighting image of the current environment using a diffusion model based on the current environment image and the lighting requirement.
[0035] The target lighting image of the current environment is an image under an ideal lighting state calculated by the diffusion model based on the current environment image and the lighting demand. The ambient brightness in the target lighting image of the current environment will not be too high or too low, and the brightness of the street lamps can be controlled to achieve the brightness effect shown in the target lighting image of the current environment.
[0036] The input of the diffusion model is the current environment image and the lighting requirement, and the output of the diffusion model is the target lighting image of the current environment.
[0037] Diffusion models generate new samples by simulating the diffusion and denoising process of data. Diffusion models are inspired by the diffusion process in physics, that is, how to transform structured data into unstructured noise by gradually adding noise, and then reverse the process to recover the original data. Diffusion models can generate data based on given conditions. In the generation of lighting images, the current environment image and lighting requirements can be used as conditional inputs, and the model will generate a target lighting image that meets specific lighting requirements based on these conditions.
[0038] Diffusion Models can generate target lighting images of the current environment because they can effectively capture the correlation and complex distribution between data points when simulating the diffusion process between data points. In this case, data points can be understood as a simulation of light propagation in the environment, while the diffusion process is similar to the propagation and reflection of light in the environment.
[0039] By inputting the current environment image and lighting requirements, the diffusion model can calculate the image under the ideal lighting condition based on this information. In this process, the model takes into account the brightness changes at various locations in the environment and the complexity of light propagation, thereby generating a target lighting image. This target lighting image will keep the ambient brightness within a suitable range, neither too high nor too low, meeting people's needs for comfortable lighting.
[0040] The brightness of the street lamp can be adjusted according to the brightness effect shown in the target lighting image, thereby adjusting the lighting of the environment so that the overall lighting effect is closer to an ideal state.
[0041] The diffusion model can effectively generate a target lighting image of the current environment, helping to achieve a more comfortable and appropriate lighting effect.
[0042] Step S5, determining the light brightness of each streetlight in the current environment image using a light brightness determination model based on the current environment image, the target lighting image of the current environment, and multiple streetlight positions in the current environment image.
[0043] The light brightness determination model is a convolutional neural network model. The input of the light brightness determination model is the current environment image, the target lighting image of the current environment, and the positions of multiple lighting street lamps in the current environment image, and the output of the light brightness determination model is the light brightness of each lighting street lamp in the current environment image.
[0044] A convolutional neural network (CNN) may be a multi-layer neural network (e.g., including at least two layers). The at least two layers may include at least one of a convolutional layer (CONV), a rectified linear unit (ReLU) layer, a pooling layer (POOL), or a fully connected layer (FC). A convolutional neural network can extract useful features from an image and gradually understand and learn contextual information of the image.
[0045] Convolutional neural networks can capture spatial relationships in input data, including position information between pixels. In this case, the convolutional neural network can use the location information of the lighting street lamps, combined with the spatial structure of the current environment and the target lighting image, to infer the light brightness that each lighting street lamp should have. This position awareness helps the model to more accurately predict the light brightness of each lighting street lamp.
[0046] Convolutional neural networks have excellent feature extraction capabilities when processing image data. Through multi-layer convolution and pooling operations, convolutional neural networks can learn various features in images, including edges, textures, shapes, etc. These features are crucial for understanding ambient lighting conditions and lighting requirements.
[0047] By taking the current environment image, target lighting image and street lamp position as input, and training the entire network for end-to-end learning, the convolutional neural network can directly output the brightness of each street lamp. This end-to-end learning method allows the model to learn complex features and rules from raw data without manually designing complex feature extraction and preprocessing processes.
[0048] Directly using the current environment image to determine the light brightness may result in unsatisfactory lighting effects, because the current environment image does not take into account the actual lighting requirements. The target lighting image generated by the diffusion model can simulate the ideal lighting state, which helps to optimize the lighting effect and ensure that the environment is neither too bright nor too dark.
[0049] The diffusion model can generate high-resolution and detailed target lighting images, which provides more accurate input for the light brightness determination model. This allows for more precise adjustments to the brightness of each street light to achieve the desired lighting effect. By simulating the propagation and reflection of light in the environment, the diffusion model can better understand the distribution of light in the environment and adjust the lighting settings in a targeted manner.
[0050] The brightness of each street lamp is determined by a convolutional neural network model, avoiding the situation where blindly increasing the brightness of all street lamps leads to excessive energy consumption.
[0051] Step S6: controlling each streetlight in the current environment image based on the light brightness of each streetlight in the current environment image.
[0052] In one embodiment, the light brightness of each street lamp in the current environment image may be sent to a street lamp control system, and the street lamp control system sends a control instruction to control each street lamp in the current environment image.
[0053] In some embodiments, a monitoring image after the lighting street lamp is lit can also be obtained, and then the similarity between the monitoring image after the lighting street lamp is lit and the current environment image is calculated. If the similarity is less than a set threshold, it is determined that the light brightness setting of each lighting street lamp in the current environment image is correct. In one embodiment, the method for calculating the similarity between the monitoring image after the lighting street lamp is lit and the current environment image is to calculate the similarity through the color histogram of the monitoring image after the lighting street lamp is lit and the current environment image.
[0054] Based on the same inventive concept, Figure 2 A schematic diagram of a control system for a street lamp on a municipal road provided by an embodiment of the present invention, wherein the control system for the street lamp comprises: An acquisition module 21 is used to acquire surveillance videos around street lamps; An environmental information processing module 22 is used to determine the current environmental image, the positions of multiple lighting street lamps in the current environmental image, the number of vehicles, the speed of vehicles, the number of pedestrians, and the speed of pedestrians based on the monitoring video around the street lamp using an environmental information processing model; A lighting demand determination module 23, configured to determine the lighting demand based on the number of vehicles, the number of people, the speed of the vehicles, and the speed of the people; A target lighting image generating module 24 is used to determine a target lighting image of the current environment using a diffusion model based on the current environment image and the lighting requirement; A light brightness determination module 25, configured to determine the light brightness of each streetlight in the current environment image using a light brightness determination model based on the current environment image, the target lighting image of the current environment, and the positions of multiple streetlights in the current environment image; The control module 26 is used to control each streetlight in the current environment image based on the light brightness of each streetlight in the current environment image.
Claims
1. A method for controlling streetlights on municipal roads, characterized in that: include: Get surveillance video around street lights; Determine the current environment image, the positions of multiple lighting street lamps in the current environment image, the number of vehicles, the speed of vehicles, the number of people, and the speed of people using an environment information processing model based on the surveillance video around the street lamp; Determine the lighting demand based on the number of vehicles, the number of people, the speed of vehicles, and the speed of people; Determine a target lighting image of the current environment using a diffusion model based on the current environment image and the lighting demand; Determine the light brightness of each streetlight in the current environment image using a light brightness determination model based on the current environment image, the target lighting image of the current environment, and multiple streetlight positions in the current environment image; Each streetlight in the current environment image is controlled based on the light brightness of each streetlight in the current environment image.
2. The method for controlling a street lamp according to claim 1, characterized in that: Determining the lighting demand based on the number of vehicles, the number of pedestrians, the vehicle speed, and the pedestrian speed includes: performing weighted summation of the number of vehicles, the number of pedestrians, the vehicle speed, and the pedestrian speed according to a preset weight coefficient to obtain the lighting demand.
3. The street lamp control method according to claim 1, characterized in that: The controlling of each street lamp in the current environment image based on the light brightness of each street lamp in the current environment image includes: sending the light brightness of each street lamp in the current environment image to a street lamp control system, and the street lamp control system sends a control instruction to control each street lamp in the current environment image.
4. The method for controlling a street lamp according to claim 1, characterized in that: The input of the diffusion model is the current environment image and the lighting requirement, and the output of the diffusion model is the target lighting image of the current environment.
5. The method for controlling a street lamp according to claim 1, characterized in that: The environmental information processing model is a recurrent neural network model.
6. A control system for street lights on municipal roads, characterized in that: include: An acquisition module is used to acquire surveillance videos around street lamps; An environmental information processing module, used to determine a current environmental image, positions of multiple lighting street lamps in the current environmental image, the number of vehicles, the speed of vehicles, the number of pedestrians, and the speed of pedestrians based on the surveillance video around the street lamp using an environmental information processing model; A lighting demand determination module, configured to determine the lighting demand based on the number of vehicles, the number of people, the speed of the vehicles, and the speed of the people; A target lighting image generation module, configured to determine a target lighting image of the current environment using a diffusion model based on the current environment image and the lighting requirement; A light brightness determination module, configured to determine the light brightness of each streetlight in the current environment image using a light brightness determination model based on the current environment image, the target lighting image of the current environment, and the positions of multiple streetlights in the current environment image; A control module is used to control each streetlight in the current environment image based on the light brightness of each streetlight in the current environment image.
7. The street lamp control system according to claim 6, characterized in that: The lighting demand determination module is also used to: perform weighted summation of the number of vehicles, the number of pedestrians, the speed of vehicles, and the speed of pedestrians according to a preset weight coefficient to obtain the lighting demand.
8. The street lamp control system according to claim 6, characterized in that: The control module is also used to send the light brightness of each lighting street lamp in the current environment image to the street lamp control system, and the street lamp control system sends a control instruction to control each lighting street lamp in the current environment image.
9. The street lamp control system according to claim 6, characterized in that: The input of the diffusion model is the current environment image and the lighting requirement, and the output of the diffusion model is the target lighting image of the current environment.
10. The street lamp control system according to claim 6, characterized in that: The environmental information processing model is a recurrent neural network model.