The invention discloses an intelligent traffic signal lamp control method based on video monitoring, and the method comprises the following steps: S1, collecting the video data of a traffic intersection in real time, the resolution being 720p, and the frame rate being 30fps; the method comprises the following steps: S1, extracting traffic parameters of a time sequence from video data by using an image recognition technology, S3, carrying out time sequence analysis by using a long short-term memory network, S4, inputting real-time and predicted traffic features into an isolated forest iForest algorithm to carry out abnormal congestion detection, S5, designing a reinforcement learning environment, S6, applying a Q-1 earni ng algorithm to carry out abnormal congestion detection, and S5, carrying out abnormal congestion detection. The optimal control strategy Pi output by the algorithm adjusts the control parameters of the signal lamp in real time, the adjustment frequency is once every five minutes, video data of a traffic intersection are collected in real time through a video monitoring technology, accurate data support is provided for traffic signal lamp control, and the traffic flow and speed in a future time period are predicted through an LSTM model. And the possible traffic jam condition is timely identified through iForest, and corresponding countermeasures are triggered.
本发明公开了一种基于视频监控的智能交通信号灯控制方法,包括以下步骤:S1实时采集交通路口的视频数据,分辨率为720p,帧率为30fps;S2利用图像识别技术从视频数据中提取时间序列的交通参数,S3利用长短期记忆网络进行时间序列分析,S4将实时和预测的交通特征输入到孤立森林iForest算法中进行异常拥堵检测,S5设计一个强化学习环境,S6应用Q‑l earni ng算法,算法输出的最优控制策略π实时调整信号灯的控制参数,调整频率为每5分钟一次,通过视频监控技术,实时采集交通路口的视频数据,为交通信号灯控制提供了准确的数据支持,通过LSTM模型预测未来时间段的交通流量和速度。通过iForest及时识别可能的交通拥堵情况,并触发相应的应对措施。
Intelligent traffic signal lamp control method based on video monitoring
一种基于视频监控的智能交通信号灯控制方法
2024-11-22
Patent
Electronic Resource
Chinese
Traffic signal lamp intelligent control method, storage medium and traffic signal lamp
European Patent Office | 2020
|Intelligent signal lamp system based on video monitoring
European Patent Office | 2023
|