基于SVM的人体运动分析与防跌倒检测技术  

Human motion analysis and fall prevention detection based on SVM

作　　者：任宇飞[1] REN Yu-fei(Shaanxi College of Communications Technology, Xi’an 710018, China)

机构地区：[1]陕西交通职业技术学院

出　　处：《信息技术》2019年第6期56-59,68,共5页Information Technology

基　　金：西安社会科学规划基金(18T02)

摘　　要：随着人们对健康监测关注度的日益增长和智能手机的普及,文中提出了一种基于SVM的人体运动分析方法,并基于该方法和Android平台设计了一种防跌倒系统。通过采集智能手机内置传感器的数据,并对原始数据进行滤波去噪、加窗分割、时频特征提取及特征降维等处理,使用SVM分析与识别不同的行为。该防跌倒系统,在检测到用户处于跌倒状态时会发出求救信息。实验与测试结果表明,所提出的方法能获得97. 5%的敏感性和98. 3%的特异性,对跌倒行为具有较高的识别精度。Focusing on growing concern on health monitoring and the popularity of smartphones,a human movement analysis method based on SVM is proposed and a fall prevention Android platform system based on this method is designed. By collecting the data through built-in sensors of smartphones,and processing the original data,such as filtering and denoising,window segmentation,time-frequency feature extraction and feature dimensionality reduction,and using SVM to analyze and identify different behaviors. The fall prevention system sends a distress message when it detects that the user is in a falling state. Experiments and test results show that the proposed method can obtain sensitivity of 97. 5% and specificity of 98. 3%,has high recognition accuracy for falls.

关 键 词：人体运动分析 跌倒检测 ANDROID SVM 传感器 

分 类 号：TP216.3[自动化与计算机技术—检测技术与自动化装置]