A window-based image processing technique for quantitative and qualitative analysis of road traffic parameters

Abstract

Traffic information is an important tool in the planning, maintenance, and control of any modern transport system. Of special interest to traffic engineers are parameters of traffic flow such as volume, speed, type of vehicle, queue parameters, traffic movements at junctions, etc. Various algorithms, mainly based on background differencing techniques, have been applied for this purpose. Since background-based algorithms are very sensitive to ambient lighting conditions, they have not yielded the expectative results. In this paper, we describe a novel approach to measure traffic parameters. This approach is based on applying edge-detection techniques to the key regions or windows. This method of measuring road traffic parameters eliminates the need of a background frame, which is an essential, but unreliable technique for background-based image-detection methods. A dynamic threshold selection technique has also been introduced to select the threshold value automatically. The image process algorithm has been applied to measure basic traffic parameters such a traffic volume, types of vehicles, as well as the complex traffic parameters such as queue parameters and movements of vehicles at a traffic junction.