Motion recognition based on a three-axis accelerometer

Elderly people who are living alone may suffer an amount of security risks in daily life. Therefore, it is obviously important for elderly people to conduct the motion recognition and the emergent situation detection with an alarm. This paper discusses a novel method of body motion classification using horizontal and vertical acceleration variance as a threshold. The signal from a three-axis accelerometer has been analyzed with a time window of 0.5 second. Every 0.5 second data corresponds to a specific classification result. The experimental results show that this method has high recognition accuracy, and it can be used for classification of most daily actions.