Abstract
This paper describes an anomaly behavior detection system for elderly people based on laser range scanner with rotary drive. This system can detect the human surface coordinate points, which is useful for the anomaly behavior detection such as falling down and montionless. Some feature values such as the center of gravity and the motion amount are extracted from the obtained data and the anomaly behavior patterns are detected by One-Class SVM. Some experiments are performed to verify the effectiveness of the proposed method.
