Real-time security solution for automatic detection and tracking of intrusion

This paper presents an effective system for the detection and tracking of intrusion in a secure space with the use of digital image processing techniques and embedded systems. The proposed method is able to identify intrusion, send alerts and then track the object of interest. First an adaptive background subtraction model has been suggested for the detection of intrusion. This adaptive method uses varying threshold for effective detection in day and night time and constantly updates the background model to account for changes due to noise. Effective detection is followed by notifying the concerned authorities via SMS and email. The detected object is then tracked using a combination of three frame differencing and FAST feature detection algorithm. This system has been suggested as a security solution for intruder detection in real-time. Experimental results indicate that the proposed solution is both feasible and effective.