Vehicle counting and classification for traffic surveillance using Wireless Video Sensor Networks

This paper describes a technique for counting and classifying vehicles in traffic video surveillance applications. We concentrate on systems requiring rapid deployment built using Wireless Video Sensor Networks. To achieve this, we propose an algorithm that employs a pixel-based Gaussian background modeling technique. The goal is to classify vehicles based on structural analysis of foreground objects. Discrimination relies on the size of the minimum bounding rectangle that encloses each foreground object identified as vehicle. The algorithm was designed to operate in resource-constrained environments.