Abstract

Info

Author(s):: WU Qiuwen; LIAO Wei; School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, Shanghai 201620, China

Keywords:: Keywords: wearable device human body area communication channel attenuation pose recognition

Abstract:: Abstract: A pose recognition method using wearable devices and human body area communication technology is proposed to overcome the drawbacks of current methods based on image and millimeter wave radar and realize the recognition through the channel attenuation caused by human posture. Five postures of human standing, squatting, sitting, running and walking are modeled, and two channels are constructed with one transmitter and two receivers for calculating the channel attenuation with 5 postures as the data set. The highest recognition accuracy of multiple single classifiers for the two channels is 94.33%, and the accuracy is improved to 99.67% with feature fusion of the two channels using Stacking method. The results show that the pose recognition method using wearable devices and human area communication technology is feasible, and that the highest accuracy rate using Stacking feature fusion is improved by 5.34% as compared with single feature.