Efficient vehicle detection and classification for traffic surveillance system

Abstract

Video surveillance systems is a key component of any security system. Making an intelligent system that can detect and track multiple moving objects from video and also deals with dynamic backgrounds, illumination problem and environment conditions is a challenging task. The proposed system is designed for real-time vehicle detection and classification. The traffic is increasing day by day due to increase in number of vehicles. Vehicle detection, classification, and counting is a very important application by which highway monitoring, traffic planning, analysis of the traffic flow, etc. can be easily done. In this paper, vehicle detection is done by background subtraction and from each detected vehicles Scale-Invariant Feature Transform (SIFT) features is extracted. Vehicles are classified using the neural network and Support Vector Machine (SVM). SVM showed better generalization than Artificial Neural Networks.