基于时空图的夜间车流量检测算法

Night traffic flow detection algorithm based on space-time diagram

夜间由于道路光线不好,不容易检测出车流量。针对这个问题,本文提出了一种基于时空图夜间车流量检测的有效方法。该方法首先将道路交通监控视频拍摄到的的图像,利用时空图转换为时间-空间坐标连续图。然后通过时空图提取到的车灯信息,采用车灯配对的方法来识别车辆,并计算出一定时间内的车流量。本实验所用的监控视频图像由架在路边高层建筑上的摄像机拍摄所得。实验结果表明:该检测方法鲁棒性较强,能够准确快速的检测出一定时间内的夜间车流量,检测准确率高达97.7%,其实用价值非常的高。

It is not easy to detect the traffic flow at night because the road light is not good.In order to solve this problem,In this paper,an effective method of night traffic flow detection based on space-time diagram is proposed.Firstly,this method converts the road traffic monitoring video image into a timespace coordinate continuous graph by using the space-time diagram.Then the vehicle light information is extracted through the space-time diagram,The vehicle light matching method is adopted to identify the vehicle and calculate the traffic flow in a certain period of time.The monitoring video used in this experiment was captured by a camera mounted on a roadside high-rise building.The experimental results showed that,the method is robust andaccurately and quickly detect the night traffic flow within a certain time.The detection with considerable practical value accuracy was as high as 97.7%.