Robust image-based detection of activity for traffic control

Abstract

An important application of image processing and computer vision is the development of intelligent systems for traffic monitoring, management and optimization. This paper presents a system for real-time detection of moving vehicles approaching an intersection from sequences of colour images acquired by a stationary camera. The proposed approach is developed in the context of traffic-light control systems. As the system is dedicated to outdoor applications, efficient and robust vehicle detection under various weather and illumination conditions must be achieved. To deal with these ever-changing conditions, the vehicle detection relies on motion segmentation with dynamic background representation and on an original switching algorithm using hue-saturation-value (HSV) colour mapping to achieve feature space segmentation. Experimental results using real outdoor image sequences demonstrate the systems robustness under various and difficult environmental conditions.