计算机工程 ›› 2019, Vol. 45 ›› Issue (2): 178-183.doi: 10.19678/j.issn.1000-3428.0048883

• 人工智能及识别技术 • 上一篇    下一篇

基于智能手机的跌倒行为识别算法研究

杨晨晨,马春梅,朱金奇   

  1. 天津师范大学 计算机与信息工程学院,天津 300387
  • 收稿日期:2017-10-09 出版日期:2019-02-15 发布日期:2019-02-15
  • 作者简介:杨晨晨(1995—),女,学士,主研方向为群智感知、移动计算;马春梅(通信作者),讲师、博士;朱金奇,副教授、博士。
  • 基金项目:

    天津市自然科学基金(17JCYBJC16400,18JCQNJC70200,18JCYBJC85900);天津大创计划项目(201710065075);天津师范大学博士基金(043-135202XB1615)

Study of Fall Behavior Identification Algorithm Based on Smartphone

YANG Chenchen,MA Chunmei,ZHU Jinqi   

  1. School of Computer and Information Engineering,Tianjin Normal University,Tianjin 300387,China
  • Received:2017-10-09 Online:2019-02-15 Published:2019-02-15

摘要:

利用智能手机的感知和计算能力,对跌倒行为识别算法进行研究。分析使用多特征传感器识别跌倒行为的必要性,并根据传感器对跌倒行为敏感程度的不同,提出基于主成分分析的多特征选择方法。针对传统K-means算法不能反映数据分布差异的问题,设计基于相对熵的跌倒行为识别算法,利用数据集分布距离进行跌倒行为识别。在真实环境下采集跌倒行为数据对算法性能进行评估,结果表明该算法能较好地识别跌倒行为,识别准确率高达96.7%。

关键词: 智能手机, 跌倒行为识别, 多特征选择, 主成分分析, 相对熵

Abstract:

This paper uses the perception and computing capacity of smartphones to study the fall behavior recognition algorithm.It analyzes the necessity of using multi-feature sensors to detect the fall behavior.According to the sensitivity of various sensors to the fall behavior,a multi-features selection method based on Principal Component Analysis(PCA) is proposed.Aiming at the problem that traditional K-means algorithm can not reflect the difference of distribution of data,a relative entropy based behavior identification algorithm is proposed.It uses the distance of distribution of data for fall behavior detection.It collects the data of the fall behavior under realistic environments and evaluates the performance of the proposed algorithm.Results show that the proposed algorithm can effectively detect fall behaviors and the accuracy is up to 96.7%.

Key words: smartphone, fall behavior identification, multi-feature selection, Principal Component Analysis(PCA), relative entropy

中图分类号: