| 摘 要: 交通阻塞通常是由于城市路口实际通行能力不足造成的，针对这一现状，设计了基于计算机视觉和深度学习的流量监控智能交通灯系统。近年来，计算机视觉以及神经网络技术[1]越来越成熟，对车辆以及行人的识别与检测[2]越来越准确。本文模拟实际的道路交通路口，使用树莓派[3]为主要控制器搭建模拟路口场景，将摄像头采集的数据通过图像处理以及神经网络，确定行车道路以及行人道路上的车辆数目和行人数目。通过数学建模确定交通灯的时延，动态设定交通灯亮灭的时长，彻底改变传统的交通灯控制模式，从而有效地缓解交通阻塞。结果表明，本系统根据车辆和行人数目动态优化交通灯时延，达到了实现交通灯智能化的目的。 |
| 关键词: 图像处理；神经网络；树莓派；目标识别 |
|
中图分类号: TP391
文献标识码: A
|
|
| Design and Implementation of Intelligent Traffic Light System for Traffic Monitoring based on Computer Vision |
|
QIN Xiaohui
|
(School of Computer Engineer, Taiyuan institute of technology, Taiyuan 030008, China)
qinxh@tit.edu.cn
|
| Abstract: Traffic jams are usually caused by insufficient actual traffic capacity at urban intersections. Aiming at this problem, this paper proposes to design an intelligent traffic light system for traffic monitoring based on computer vision and deep learning. In recent years, with the increasing maturity of computer vision and neural network technology[1], recognition and detection[2] of vehicles and pedestrians have become more and more accurate. Actual road traffic intersections are simulated in this paper, and Raspberry Pi[3] is used as the main controller to build the simulated intersection scene. Data collected by camera determines the number of vehicles and pedestrians on driving and pedestrian roads through image processing and neural network. Delay of the traffic light is determined through mathematical modeling and duration of traffic lights on and off is dynamically set. Thus, traditional traffic light control mode is completely changed, so to effectively alleviate traffic jams. Results show that the system dynamically optimizes traffic lights delay according to the number of vehicles and pedestrians, and achieves the goal of intellectualizing traffic lights. |
| Keywords: image processing; neural network; Raspberry Pi; target recognition |
