摘要
低成本、易实施、无接触式的步态信号采集与分类平台可为步态的定量评估、智能诊断提供新的研究路径与参考思路。应用改进经验模态分解(empirical mode decomposition,EMD)结合Bagging正则化共空间模式(Bagging regularized common spatial pattern,BRCSP)与Fisher线性判别分析法(Fisher linear discriminant analysis,FLDA)设计了步态信号采集与分类平台。该平台将采集到的步态信号通过改进EMD进行滤波,抑制高频干扰噪声,提取真实、有用的本征模态分量(intrinsic mode function,IMF)进行信号重构,从而得到包含完整、准确步态信息的信号;再通过BRCSP特征提取方法强化步态信号的个体差异性及共同特征,提取出显著的特征分量;最后应用FLDA方法将特征向量映射到低维度空间中并进行步态分类。实验结果表明,该平台能准确分类进入红外、激光检测区域内的不同步态,平均分类准确率达到96.6%。
Low cost,easy implementation and non-contact gait signal acquisition and classification platform can provide a new research path and reference ideas for quantitative evaluation and intelligent diagnosis of gait.A gait signal acquisition and classification platform is designed by using improved empirical mode decomposition(EMD)combined with Bagging regularized common spatial pattern(BRCSP)and Fisher linear discriminant analysis(FLDA).The collected gait signals are filtered by improved EMD to suppress high-frequency noise and extract the real and useful intrinsic mode function(IMF)for signal reconstruction,thus obtaining complete and accurate gait signals by this platform.BRCSP feature extraction method is used to enhance individual differences and common features of gait signals,and significant feature components are extracted.The feature vector is projected into the lower spaces with FLDA and then classify the gait.Experimental results show that the proposed method can accurately classify different gaits in infrared and laser detection regions,with an average classification accuracy of 96.6%.
作者
陈东毅
李玉榕
CHEN Dongyi;LI Yurong(College of Electrical Engineering and Automation,Fuzhou University,Fuzhou 350108,China;Fujian Key Lab of Medical Institute and Pharmaceutical Technology,Fuzhou University,Fuzhou 350108,China)
出处
《贵州大学学报（自然科学版）》
2022年第4期60-66,共7页
Journal of Guizhou University:Natural Sciences
基金
国家自然科学基金资助项目(61773124)
福建省教育厅科研资助项目(JAT170109)。
