摘要
为降低数据噪声的干扰,提高步态识别的有效性,提出一种基于隐马尔可夫模型的步态识别算法(GR-HMM)。利用滑动平均滤波模型对步态数据进行去噪预处理,建立观测序列;基于GR-HMM算法计算观测序列概率,重估迭代至观测序列概率收敛,得到GR-HMM算法的参数模型;对步态数据进行步态阶段识别及步态关键事件定位。基于Shimmer IMU采集的数据对算法进行训练和分析评估,实验结果表明,GR-HMM算法的步态阶段识别灵敏度和特异性分别可达93.1%和96.9%。
To reduce the interference of data noise and improve the effectiveness of gait recognition,agait recognition algorithm based on hidden Markov model(GR-HMM)was proposed.The moving average filter model was used to pre-process the gait data,and the observation sequence was established.The probability of the observed sequence was calculated based on the GRHMM algorithm,and the parameter model of GR-HMM algorithm was obtained by iterating to the probability convergence of observation sequence.Gait phase identification and gait key event location were carried out for gait data.The algorithm was trained and evaluated based on the data collected by Shimmer IMU.Experimental results show that the sensitivity and specificity of GR-HMM algorithm for gait phase recognition can reach 93.1% and 96.9%respectively.
作者
刘畅
魏忠诚
张春华
王巍
赵继军
LIU Chang;WEI Zhong-cheng;ZHANG Chun-hua;WANG Wei;ZHAO Ji-jun(School of Information and Electrical Engineering,Hebei University of Engineering,Handan 056038,China;Hebei Key Laboratory of Security and Protection Information Sensing and Processing,Hebei University of Engineering,Handan 056038,China;Department of Public Sports,Hebei University of Engineering,Handan 056038,China)
出处
《计算机工程与设计》
北大核心
2019年第12期3487-3493,共7页
Computer Engineering and Design
基金
国家自然科学基金项目(61802017)
河北省自然科学基金项目(F2018402251)
邯郸市科学技术研究与发展计划基金项目(1721203048)
河北省物联网数据采集与处理工程技术研究中心开放课题基金项目(2016-2)
