Teerasak Kroputaponchai and Nikom Suvonvorn, (2009), Vision-based Fall Detection and Alert System Suitable for the Elderly and Disable Peoples, In Proceedings of 24th Japanese Conference on the Advancement of Assistive and Rehabilitation Technology (JCAART), August 26-28, 2009, Tokorozawa, Japan, p.217-218.
Automatic monitoring of the activities of daily living (ADL) for elderly and disabled people from image sequences is an important issue for homecare or care centre. Falls amongst the elderly are particularly serious and often lead to injury or death . This paper proposes a fall detection algorithm with alert system based on video analysis techniques. In the first part, the fall detection algorithm is described: the mixture motion-texture model for human detection and tracking is firstly explored, and then the human’s body characteristics are extracted and defined as robust features. The Spatial-Temporal Features is specially constructed in order to detect the fall events. Our detection algorithm is build based on this representation.
The second part emphasizes on the alert system. A central system for fall analysis framework is designed to deal with the heavy processing for large amount of video data from multiple cameras that needs to be executed in real-time with supported multi-cores multi-processors architecture and resource management capacity such as CPU, memory, and network bandwidth. In the alert process, the detected events obtained by framework may be sent to the map-based alert service that can be remotely monitored by the experts or the correspondent online call center.