基于机器视觉的交叉口特种车辆快速通行技术研究

交叉口是道路网络中重要的交通节点，容易产生交通堵塞问题，为了在保证通行安全的情况下提高特种车辆的通行效率，研究基于机器视觉的交叉口特种车辆快速通行技术。优化通行基础采用图像采集及预处理、检测识别和通行控制作为技术框架结构，利用机器视觉技术采集交叉口实时交通图像，通过图像滤波、图像增强等步骤，实现初始图像的预处理。利用Car-YOLO网络识别交叉口通行能力，规划快速通行路线，考虑前车行驶状态，求解特种车辆通行速度，针对车辆所占车道，通过绿灯早启、绿灯周期时间延长等方式控制交叉口信号灯，实现交叉口特种车辆快速通行。实验结果表明:在拥堵和正常通行场景下，优化设计技术的特种车辆通过时间的平均值分别为18.2s、10.1s，事故发生概率分别低于2%、1.4%，具有较好的应用效果。

Research on Rapid Transit Technology of Special Vehicles at Intersections Based on Machine Vision

Intersection is an important traffic node in the road network, which is prone to traffic congestion. In order to improve the efficiency of special vehicles while ensuring traffic safety, a machine vision based rapid traffic technology for special vehicles at intersections is studied. The optimization of traffic foundation adopts image acquisition and preprocessing, detection and recognition, and traffic control as the technical framework structure. Machine vision technology is used to collect real-time traffic images at intersections, and initial image preprocessing is achieved through image filtering, image enhancement, and other steps. Using the Car-YOLO network to identify the traffic capacity of intersections, plan fast traffic routes, consider the driving status of the preceding vehicle, calculate the speed of special vehicles, and control the intersection signal lights based on the lane occupied by vehicles, such as early green light and extended green light cycle time, to achieve rapid passage of special vehicles at intersections. The experimental results show that in congested and normal traffic scenarios, the average passing time of special vehicles with optimized design technology is 18.2 seconds and 10.1 seconds, respectively, and the probability of accidents is less than 2% and 1.4%, respectively. It has good application effects.