Analysis and monitoring of a high density traffic flow at T-intersection using statistical computer vision based approach

摘要

A reliable traffic flow monitoring and traffic analysis approach using computer vision techniques has been proposed in this paper. The exponential increase in traffic density at urban intersections in the past few decades has raised precious and challenging demands to computer vision algorithms and technological solutions. The focus of this paper is to suggest a statistical based approach to determine the traffic parameters at heavily crowded urban intersections. The algorithm in addition to accurate tracking and counting of freeway traffic also offers high efficiency for determining vehicle count at a high traffic density T-intersection. The system uses Intel Open CV library for image processing. The implementation of algorithm is done using C++. The real time video sequence is obtained from a stationary camera placed atop a high building overlooking the particular T intersection. This paper suggests a dynamic method where each vehicle at a T intersection is passed through a number of detection zones and the final count of vehicles is derived from a statistical equation.