电子学报 ›› 2017, Vol. 45 ›› Issue (3): 570-576.DOI: 10.3969/j.issn.0372-2112.2017.03.010

• 学术论文 • 上一篇    下一篇

基于三轴传感器的老年人日常活动识别

汪成亮1,2, 王小均2   

  1. 1. 重庆大学信息物理社会可信服务计算教育部重点实验室, 重庆 400044;
    2. 重庆大学计算机学院, 重庆 400044
  • 收稿日期:2015-07-06 修回日期:2015-11-24 出版日期:2017-03-25
    • 通讯作者:
    • 汪成亮
    • 作者简介:
    • 王小均 男,1991年1月生于重庆.硕士.主要研究方向为物联网、穿戴式计算.E-mail:20095358@cqu.edu.cn
    • 基金资助:
    • 国家自然科学基金资助项目 (No.61004112); 中央高校基本科研基金资助项目 (No.CDJZR12180006)

Daily Activity Recognition Based on Triaxial Accelerometer of Elderly People

WANG Cheng-liang1,2, WANG Xiao-jun2   

  1. 1. Key Laboratory of Dependable Service Computing in Cyber Physical Society, Ministry of Education College of Computer Science, Chongqing University, Chongqing 400044, China;
    2. Computer School, Chongqing University, Chongqing 400044, China
  • Received:2015-07-06 Revised:2015-11-24 Online:2017-03-25 Published:2017-03-25
    • Supported by:
    • National Natural Science Foundation of China (No.61004112); Fundamental Research Funds for the Central Universities (No.CDJZR12180006)

摘要:

本文针对老年人日常活动类型及特点提出了一种基于三轴加速度传感器和HMM(Hidden Markov Model)的活动识别方法.本文首先提取了针对老年人相异、相似活动的标准差、能量、相关系数、RAF(RAtio Forward)、RVF(Ratio Vertical Forward)等特征值.然后定义老年人的HMM活动识别模型.最后在经过Baum-Welch算法对HMM进行参数训练后使用Viterbi算法来进行老年人活动识别.实验结果表明,本文方法适用于老年人的日常活动的识别,平均识别精度达到了93.3%,尤其是对于相似步态活动的识别准确率达到了93.7%.

关键词: 活动识别, 三轴传感器, 特征提取, 隐马尔科夫模型

Abstract:

In the light of the motion type and characteristics of elderly people,we propose an approach which is based on triaxial accelerometer and hidden Markov model(HMM) for activities recognition.Firstly,we extract standard deviation(SD),energy,correlation coefficients,ratio forward(RAF),ratio vertical forward(RVF) as the features corresponding to different and similar activities of elderly people.Secondly,we define the activities recognition model based on HMM for elderly people.Finally,we use the Viterbi algorithm to recognize the activities for elderly people after the parameters are trained by Baum-Welch algorithm.The experimental results shows that our approach is can be applied for daily activity recognition of elderly people and the average recognition accuracy is 93.3%,specifically the accuracy of similar walking activities is 93.7%.

Key words: activity recognition, triaxial accelerometer, feature extract, hidden Markov model(HMM)

中图分类号: