Real-time traffic accident detection based on the derivative of traffic parameters

This paper focuses on the study of the derivative of traffic parameters for detecting traffic accidents in real time. The temporal properties of both the traffic flow and the vehicle speed are analyzed per day in May. It is found that the falling and rising of both the traffic flow and the vehicle speed have a relationship with the traffic accident happening on the urban road. Then, the traffic accident on the urban road can be detected in real time by the shape variation of falling and rising rapidly over time, namely, their derivatives. Through surveillance video and traffic accident records from the Traffic Police Detachment of Fuzhou Public Security Bureau (TDFPSB), the method based on the derivatives of both traffic flow and vehicle speed is verified to identify the traffic accident from traffic congestion.

I. Introduction

A traffic accident is a major public security problem in the global world, especially in developing countries [1]. Lots of people suffer from traffic accidents in urban cities each year [2]. If the road traffic accident is not addressed, its secondary damage and traffic congestion may happen within a short period. To avoid these harms, the traffic authorities tend to monitor road traffic and automatically send warning to drivers in Intelligent Transportation System (ITS) [3]. Also, what is crucial is how to detect traffic accidents in real time. Therefore, real-time traffic accident detection plays an important role in the warning and handling of traffic accidents.