An Efficient Approach of Integrating RFID and Vision Techniques in Object Surveillance System

With the automatically detecting and identifying ability, RFID has been widely used in object monitoring. However, due to the limited functionality of a single sensor, the RFID-based surveillance system doesn't do well in forensics and future tracking. To realize both real-time monitoring and automatically evidences extracting, a RFID and vision based object surveillance system is presented in this paper. Differing from other systems, we utilize motion detection results to control the video recording and tag reading. Then the stream processing and the complex event processing techniques are used to detect events of interest from the volume of raw RFID data in real time. Finally, an event-driven mechanism is proposed to integrate RFID and vision detected results. Also the effectiveness and efficiency of the proposed approach have been studied in this paper.