Visual based trespasser and faint detection via human silhouette integration with posture recognition

This paper proposes an efficient visual based home surveillance system for security and health care purposes. The proposed surveillance system is used to detect trespasser or fainted individuals in an indoor location such as private property, home and old folk's center. In this system, a wireless webcam is used to capture the required scene then the scene is fed via a wireless router into a personal computer for image processing, trespasser or faint detection and alarm activation. The human silhouette integration includes the usage of head, leg and ratio detection features for a more robust application. Experimental results show that the proposed surveillance system, in an indoor environment with a good light condition achieves high accuracy of 92.41% in trespasser detection and 98.44% in faint detection