基于智能手机的步态识别研究与设计

Research and Design of Gait Recognition Based on Smartphone

在当前大数据浪潮的影响下,生物识别技术的发展吸引了越来越多国内外研究者的关注。步态识别技术能够根据受试者行走时的步态,进行非接触式身份识别,具有采集方便、识别距离远、不易隐藏等优点。文章提出了一种基于智能手机的步态识别方法,选用加速度、陀螺仪传感器采集步态数据,利用机器学习技术对步态信息进行分析处理,以参数调优后的随机森林算法作为步态识别分类模型,经测试模型准确率可达94.05%。此外,设计开发了一款实时的智能手机身份识别系统,将提出的步态识别最佳分类模型部署到移动端,生成结果可用于身份识别,也为步态数据采集提供一种新的方式。

Under the influence of the current wave of big data,the development of biometric identification technology has attracted more and more attention from researchers at home and abroad.Gait recognition technology can be used for non-contact identification according to the gait of the subject when walking.It has the advantages of convenient collection,long recognition distance and not easy to hide.This paper proposes a gait recognition method based on smartphone,which selects acceleration and gyroscope sensors to collect gait data.Machine learning technology is used to analyze and process the gait information.The random forest algorithm after parameter tuning is used as the gait recognition classification model.After testing,the accuracy of the model can reach 94.05%.In addition,this paper designs and develops a real-time smartphone identity recognition system,and deploys the proposed gait recognition optimal classification model to the mobile terminal.The generated results can be used for identity recognition,and also provides a new way for gait data acquisition.