基于移动机器视觉和多维图像特征的道路交通流检测方法研究

Research on Road Traffic Flow Detection Method Based on Mobile Machine Vision and Multidimensional Image Features

本文提出一种基于移动机器视觉的多维图像特征的道路交通流检测方法,使用搭载相机的无人机在空中获取图像,再用特征提取的方法将图像降维,使用车辆跟踪算法实现车辆检测与跟踪,结合混合高斯背景建模、中值滤波等图像处理算法对图像进行处理后,通过视频中的虚拟检测线、检测框并结合相应的交通流参数算法,实现对断面流量、车速、多车道车流等交通流量检测。

A road traffic flow detection method based on multi-dimensional image features of mobile machine vision is proposed in this paper. The UAV with camera is used to acquire images in the air,and then the dimensionality of the images is reduced by feature extraction. Vehicle tracking algorithm is used to realize vehicle detection and tracking,which combines Mixed Gauss background modeling and median filtering. After image processing,virtual detection lines,detection frames and corresponding traffic flow parameter algorithms are combined to detect cross-section traffic flow,vehicle speed and multi-lane traffic flow.