A method for fast fall detection

This paper proposes a new approach to detect the fall of the elderly. The detection system is composed of a PC server and distributed cameras. The distributed cameras are used to capture images of different place in a house, and the PC server is used for fall recognition. In the fall recognition process, we extracted a new feature using for distinguishing lying down normally and behavior of falling. At the same time, the surroundings and personal information is combined into the system to makes it smarter. With the surroundings and personal information such as toilet, kitchen, bedroom, living room, personal height, personal weight, and personal electronic health history, we can adjust the detection sensitivity to reduce unnecessary alarms, and improve the systempsilas performance in detecting fall. Experimental results show that this method is practicable for fall detection.