Research on intelligent perception and human activity monitoring for people with inconvenient movement

Abstract

Aiming at the problems of various sensors, complex recognition algorithm poor implement ability and real-time performance in current human activity state recognition methods, a human activity monitor based on single three-axis acceleration sensor is designed. By collecting acceleration data of human waist, using sliding time window method to extract time domain features, four active states are identified: long-term violent active state. Long-term static state, fall state and normal active state. The technical development and the existing difficulties and problems are discussed for future related research.