Abnormality Detection and Traffic Flow Measurement Using a Hybrid Scheme of SIFT in Distributed Multi Camera Intersection Monitoring

This paper presents an unsupervised abnormality detection method using a multi camera system with clustering in real time. Among the most important research in intelligent transportation systems (ITS), automatically intersection flow monitoring is one of the critical and challenging tasks. The proposed work addresses anomaly detection by means of trajectory analysis based on single support vector machine (single-SVM) clustering. The main problem associated with vehicle tracking is the occlusion effect. Using multiple views of cameras for producing a uniform tracking configuration is more suitable for vehicle's behaviour extraction. We use a hybrid scheme of scale invariant feature transform (SIFT) to detect and recognize vehicles in multi view system, so behaviour extraction is done more accurately and conveniently. The main focus of this paper is to extract traffic flows which assists in regulating traffic lights based on smart cameras.