Real-Time Abandoned Object Detection Using Video Surveillance

Due to increase of attacks in public places, security has now become a major issue in public places. In this paper, we have proposed an abandoned object detection through video surveillance with real-time alarming. We use dual background subtraction method to find out the static object. It is been assumed that object which is part of foreground for longer period of time slowly turns as part of background. Background modelling is done using approximate median model. For foreground processing background subtracting method is followed by ANDing operation of frames to find out the static object. The system is simple to design and not having complexity of filters or complex calculation. PETS 2006 database is used for testing algorithm. The result shows satisfactory performance even under complex condition of shadow, moving crowd, and lightning condition. MATLAB R2013a is used for compilation of system.