基于BSN和神经网络的人体日常动作识别方法

Human daily activity recognition method based on BSN and neural network

基于人体传感器网络(BSN)对人体动作的识别,在远程医疗服务中具有重要应用.搭建了一个基于BSN的人体动作监测平台,实验中通过固定在人体腰部和大腿上的两个加速度传感器节点,来采集人体日常生活中的7个动作所产生的加速度信号.特征提取包含传感器节点在3个轴上信号的时域和频域信息,并采用神经网络和分层的方法融合信息对7个动作进行分类和识别.实验结果表明,应用所搭建的BSN平台和识别方法,采用两个传感器节点识别人体日常生活中的7个动作具有很高的正确率.

Human activity recognition based on body sensor networks （BSN） has broad application in remote medical service. A monitoring platform based on BSN is established, and acceleration signals of 7 human activities are collected from two sensor nodes mounted on the waist and the thigh of the user. The extractions of features include time domain and frequency domain information of tri-axis accelerometer and then 7 human daily activities are classified by using neural network and hierarchical method. The experimental results show that the established monitoring platform and proposed recognition method can achieve satisfactory performance for 7 human daily activity recognition.