Real-Time Detection of Vehicles for Advanced Traffic Signal Control

Vehicle detection by video cameras is one of the most promising new technologies for wireless large-scale data collection and implementation of advanced traffic control and management schemes such as vehicle guidance/navigation. In this paper we propose an approach to detect and count vehicles at an intersection in real-time, using a fixed camera. After identifying moving objects images via background frame differencing, edge detection, erosion and dilation operations are performed to suppress noise. Separated and rotated properly, the denoised binary image is then used to generate a vertical projection histogram from which information about the size and coordinates of each component is utilized to compute the number of vehicles. This detection algorithm gives an approximate number of vehicles. An adaptive traffic signal control strategy controls the traffic flow. The simulating results show a great efficiency of traffic control and management scheme in practice.

1. Introduction

The current Intelligent Transportation Systems (ITS) boom has caused the need for a comprehensive realtime surveillance of traffic conditions over the transportation network to take dynamic control and management of traffic into account. Parameters such as speed, flow-rate, and vehicle types, which are useful information for applications of road surveillance and traffic signal control, are measured. Current vehicle detection systems are mainly based on ultrasonic sensors [1], infrared sensors [2], inductive loops sensors [3], microwave sensors [4], and video sensors [5]. However, ultrasonic sensors which are vulnerable to pedestrian and vehicular impact, cannot provide high detection precision and the effective detection distance is no more than 12 meters [6]; infrared sensors lack of stability in noisy environments; the inductive loops sensors perform well in detection precision based on disruptive installation process and high cost on installation; the microwave sensors are unable to detect slow or stationary vehicles. In comparison, the video sensors tend to be more promising in both research and application, due to their fast response, easy installation, operation and maintenance, and their ability to monitor wide areas. [5]