Research And Application Of Wearable Computing Technology For Daily Human Activity Recongnition

Wearable computing technology for daily human activity recognition aims at perceiving and recognitizing the daily external activities of human body, using the embedded computer technology. As an embedded computer which can provide services in the form of a more active and natural services to human being, wearable technology gets more and more attention from developers and consumers. Wearable computing technology used in the information sharing and information exchange between sensors and users, is aimed at combing the computer system and human closely together. In wearable computing technology area, human activity recognition technology has become an important research direction in this field, such as intelligent home, health care, motion detection and a series of typical application. Wearable sensors is an important part of wearable computing technology, and it can provide users accurate and srable information about human daily activities.Traditional human activity recognition methods mostly use the activity perception technology based on the video monitoring. Due to the limition of monitioring scope, stong privacy invasive and more easily affected by environmental factors, more and more researchers prefer to use wearable technology for human activity recognition. However, in real life, because of the complexity and randomness of the human activity, various application about the human activity recognition has the problem that its recognition accuracy is not high, the application is relatively single and lack of power battery life.Aiming at studying the existing deficiencies of wearable human activity recognition, this paper designed a novel wearable activity perception platform based on multiple sensor for human activity recognition, and realizing the recognition of various types of human activities problem. In combination with function, power consumption, performance and other factors of system, the overall architecture of system, hardware system of wearable activity recognition platform and software system of wearable activity recognition platform were designed. Emerging Pattern(EP) based algorithm is applied to release the unification of different activity recognition algorithm framework. Though plenty of experiments and examples, 6 kinds of situation of activities are statisticed. The average recognition accuracy of human activity category is calucated as 86.2%. This results verified the feasibility and efficiency of this wearable activity perception platform.