欢迎来到天天文库
浏览记录
ID:54601789
大小:1.04 MB
页数:7页
时间:2020-05-03
《基于 GGA-Elman 网络的头部体态语言 sEMG识别-论文.pdf》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、第9卷第4期智能系统学报Vo1.9No.42014年8月CAAITransactionsOfIntellizentSystemsAgu.2014DOI:10.3969/j.issn.1673-4785.201310047网络出版地址:http://www.cnki.net/kcms/doi/10.3969/j.issn.1673—4785.201310047.html基于GGA—Elman网络的头部体态语言sEMG识别杨钟亮,陈育苗(1.东华大学机械工程学院,上海201620;2.东华大学服装·艺术设计学院,上海200051)摘要:为提高头部体态语言表达“同意”与“不同意”态
2、度的识别效果,提出结合贪心遗传算法和Elman神经网络的表面肌电识别方法。通过前导实验分别采集8名被试者点头与摇头时颈部肌肉的表面肌电信号,利用Wilcoxon秩和检验提取具有显著性差异的10个肌电时域特征值,进而基于贪心遗传算法优化的Elman神经网络建立体态语言识别模型。实验结果表明,该模型能成功识别自发表达“同意”与“不同意”的头部体态语言,与标准Elman神经网络和BP神经网络的识别模型相比,相关系数更高、均方误差更小,对测试集的正确识别率提高了3.2%以上,从而验证了该方法的可靠性。关键词:头部运动;体态语言;肌电;肌肉;时域分析;神经网络;遗传算法;模式识别中图分
3、类号:TP391文献标志码:A文章编号:1673—4785(2014)04—385—07中文引用格式:杨钟亮,陈育苗.基于GGA-Eiman网络的头部体态语言sEMG识别[J].智能系统学报,2014。9(4):385-391.英文引用格式:YANGZhon#iang。CHENYumiao.AnsEMGapproachtorecognizethebodylanguageoftheheadbasedontheGGA·Elmannetwork[J].CAAITransactionsonIntelligentSystems,2014。9(4):385-391.AnsEMGappro
4、achtorecognizethebodylanguageoftheheadbasedontheGGA.ElmannetworkYANGZhongliang,CHENYumiao(1.CollegeofMechanicalEngineering,DonghuaUniversity,Shanghai201620,China;2.Fashion·ArtDesignInstitute,DonghuaU—niversity,Shanghai200051,China)Abstract:Inordertoimprovetherecognitioneffectsofthe”agreemen
5、t”and”disagreement”attitudesexpressedbythebodylanguageoftheheadmovements,asurfaceelectromyography(sEMG)approachincombinationwiththegreedygeneticalgorithm(GGA)andtheElmanneuralnetworkisproposed.ThesEMGsignalsoftheneckmusclesweredetectedwhileeightparticipantswerenoddingandshakingtheirheadsres
6、pectivelyduringapilotexperiment.BymeansoftheWilcoxon’Ssigned—ranktest,tenfeaturesofthesEMGtimedomainindiceswereextractedwithsignificantdifferences.Furthermore,thebodylanguagerecognitionmodelwasconstructedbasedontheElmannet·workoptimizedbyGGA.Experimentalresultsshowthatthemodelcansuccessfull
7、yrecognizethe’’agreementanddisagreement”attitudesspontaneouslyexpressedbythedifferentbodylanguagesofthehead.ComparedwiththerecognitionmodelsusingthestandardElmanandBPnetwork,thecorrelationcoeficientofthispresentmodelishigher,themeansquarederrorisless,and
此文档下载收益归作者所有