SAD: Sensor-Based Anomaly Detection System for Smart Junctions

The low cost and advancement of sensor technology have long proven to be invaluable in their use in all areas of science. Several video-based methods have already been suggested for intelligent transportation systems (ITSs). However, the information collected through image/video processing alone is not sufficient for an error-proof traffic management system. In this article, we propose sensor-based anomaly detection (SAD), a system that integrates the capabilities of sensors with powerful image processing techniques to build an efficient, self-adaptive traffic control system. This system has two interlinked modules: accident detection and emergency stat invocation. The accident detection module uses piezoelectric sensors to measure any pressure variation that in turn invokes an image processing submodule to detect the accident. Once the accident is confirmed, the emergency stat invocation module contacts the nearest emergency service for help. The proposed model is simple and fast to detect anomalies in real-time heterogeneous traffic conditions. The combined use of sensor technology and image processing to detect anomalies significantly increases the accuracy of the system, i.e., by reducing false alarms. By integrating SAD in real-time, the emergency services are alerted instantly, thereby ensuring faster medical assistance. The proposed system, when deployed and analyzed in real-time, achieves 93%–95% accuracy in urban areas and 96%–99% accuracy on highways. SAD clearly offers more accuracy when compared with other state-of-the-art methods without compromising performance.