Abnormal event detection in traffic video surveillance based on local features

Abstract

In this article, we proposed an abnormal event detection method based on local features for traffic video surveillance. Firstly, foreground assumed to be moving is detected and affined with morphological operations. Then each foreground region's area, shape factors (ellipse eccentricity, width-height radio of outside rectangular, and etc.), and pixel moving velocity vector are extracted. Based on those features, regions are classified into different groups as pedestrian, vehicle or noise region, and their behavior is classified using trained local features' distribution map (location distribution and velocity distribution). Finally, a simple classifier is used to determine objects' states of normal or abnormal. With the rapid development of ITS (Intelligent Traffic Surveillance), our low complexity and low level abnormality detection method is well fit in early alarm of distributed surveillance system. We have some experiment to show the benefits of proposed method.