RVM Based Human Fall Analysis for Video Surveillance Applications

For the safety of the elderly people, developed countries need to establish new healthcare systems to ensure their safety at home. Computer vision and video surveillance provides a promising solution to analyze personal behavior and detect certain unusual events such as falls. The main fall detection problem is to recognize a fall among all the daily life activities, especially sitting down and crouching down activities which have similar characteristics to falls (especially a large vertical velocity). In this study, a method is proposed to detect falls by analyzing human shape deformation during a video sequence. In this study, Relevance Vector Machine (RVM) is used to detect the fall of an individual based on the results obtained from torso angle through skeletonization. Experimental results on benchmark datasets demonstrate that the proposed algorithm is efficient. Further it is computationally inexpensive.