3407 生体情報を用いた人間の行動判別手法の提案(S72 生物医学工学における計測と制御,S72 生物医学工学における計測と制御)

概要

論文の詳細を見る
Recently, many researches on classification of human actions have been reported. But only few simple actions can be classified, they are inadequate for modeling an entire daily life pattern. In this paper we propose a classification method of human actions using 3-dimensional acceleration sensor and electrical potential sensor data. Since we didn't know how many groups of actions compose the daily life pattern, we decided to use hierarchical clustering to plot out all key actions composing daily life. This study mainly clarified three advantages of the proposed method. Firstly, individual variability on exercise intensity was ignored by normalizing values of characteristics before discrimination. Secondly, changes of walking condition (motion velocity and body tilt) were detected. Thirdly, desk work activities were classified by combining information provided by 3-dimensional acceleration sensors and by electrical potential sensors.
一般社団法人日本機械学会の論文
2007-09-07