Design And Realization Of Daily Activity Monitoring System For Wearable Multi - Sensor Human Body

Currently, Wearable Health Monitoring is the most active areas of Wearable Computing. It has been extensively researched by many universities and companies.As micropro- cessor and sensor technology is always improving, the market emerged a large number of wearable health monitoring devices. Based on the analysis of previous researchs,this article has designed a low-power hardware solutions for Wearable Health Monitoring and also has proposed a Low-power software design modle for the hardware solutions. In order to achieve power saving,the major idea of the software design modle is to use external interrupt drive the system,put the MCU in low-power mode as much as possible.The original research of human action recognition is based on pure acceleration analysis of human action. This paper puts forward a kind of human action recognition method based on the fusion of gesture and acceleration. The human gesture can be obtained by combining acceleration with angular velocity. And then we can realize real-time decomposition of horizontal and vertical acceleration by using the gesture detection result. It makes acceleration of human action more characteristic. In this paper the new kind of human activities data analysis framework is proposed. On traditional acceleration behavior analysis, applying posture analysis, that furthers the accuracy and rationality of behavior recognition results. It uses double-deck classification model based on support vector machine (SVM).It has combined gesture with acceleration effectively and improved the practicability of the human behavior recognition. A series of experiments is conducted, which has confirmed the effectiveness of the proposed method in this paper.