Recognition Of Motion Pattern Based On Triaxial Acceleration Of Wrist

Human motion mode recognition is one of the important research fields in the field of artificial intelligence,and it is also a representative research topic in the field of recognition.The acceleration sensor has the advantages of simple equipment,low cost and high sensitivity,and the human behavior recognition based on the acceleration sensor has become a hot research topic.With the rapid development of micro-acceleration sensor in recent years,the human motion mode recognition technology based on the acceleration sensor becomes more and more mature,and has been widely used in various fields.Due to the complexity and variability of human motion itself,there are still some deficiencies in the research of the acceleration sensor,and there is a lot of space for space.In this paper,the three major sections including signal preprocessing,feature selection,extraction and identification methods in the human motion mode recognition based on acceleration sensor are deeply analyzed.The main tasks include:Firstly,the background and the present situation of human motion mode recognition are studied,and two types of behavior recognition methods based on the vision and sensor are introduced.The human behavior recognition process based on the acceleration sensor is summarized and the main modules in the systematical process are studied in detail.Secondly,the threshold denoising method based on wavelet analysis is deeply analyzed.Through the scientific simulation experiment,and compared with traditional denoising method,the high efficiency of the wavelet threshold denoising method with an improved denoising effect has been proved.The five characteristics including mean value,standard deviation,triaxial correlation coefficient,and range interquartile and wavelet energy are used for the recognition of human motion modes,and the simulation experiment is carried out with the recognition rate of more than 90%.Thirdly,dynamic time regulation(DTW)algorithm is adopted in this paper.Based on the deficiencies of traditional DTW algorithm,a semi-partial dynamic algorithm is selected for improvement,thereby realizing the recognition of five daily motion modes such as standing,walking,running and so on,and also the recognition of three confusable walking types including slow walking,normal walking and fast walking,and the recognition effect is effective and precise.Finally,taking caring of the health of old people as the starting point,this paper studies the falling behavior detection method based on DTW algorithm.The three representative characteristics of acceleration,tilt Angle and angular velocity are selected to accurately distinguish the falling behavior.