资源描述：
《2 Identifying users of portable devices with accelerometers》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、IDENTIFYINGUSERSOFPORTABLEDEVICESFROMGAITPATTERNWITHACCELEROMETERSJaniMäntyjärvi,MikkoLindholm,ElenaVildjiounaite,Satu-MarjaMäkelä,HeikkiAilistoVTTElectronics,P.O.Box1100,FI-90571Oulu,Finlandphone:+35885512111,fax:+35885512320,email:Firstname.Surname@vtt.fiABSTRACTorburdenedbythetechnologysheisusin
2、g.Gait,i.e.,walkingstyle,isfairlycharacteristicforindividuals[3],Identifyingusersofportabledevicesfromgaitsignalwhereasdeliberateimitationofotherperson'swalkingstyleacquiredwiththree-dimensionalaccelerometerswasisdifficult.studied.Threeapproaches,correlation,frequencydomainGaitrecognitionhasbeenstu
3、diedasabehavioralanddatadistributionstatisticswereused.Testsubjectsbiometricforaboutadecadeandmethodstypicallyutilized(N=36)walkedwithfast,normalandslowwalkingspeedsarebasedonmachinevision[4-7].Theperformanceoftheinenrolmentandtestsessionsonseparatedayswearingvisionbasedgaitbiometricsislowerthan,e.
4、g.,theaccelerometerdeviceontheirbelt,atback.Identifyingfingerprints,andthemethodisinitsinfancy[8].theuserswiththisnovelgaitrecognitionmethodshowedInthefieldsofmedical,wearable,andmobiletobepossible.Bestequalerrorrate(EER=7%)wascomputingtherelatedresearchmainlycoversrecognitionachievedwithsignalcorr
5、elationmethod,whilefrequencyofvarioushumanactivities,suchas,walking,running,domainmethodandtwovariationsofdatadistributionwalkingup/downthestairs/slope[9,10,11].MotivatingstatisticsmethodproducedEERof10%,18%and19%,applicationscenariosvary,e.g.,frompatientactivityrespectively.surveillanceorenergyexp
6、enditureestimationtoadaptinguserinterfacesofdevicesusersarecarrying.Thepersonidentificationapproachcanbeseenasreversaloftheresearchattemptstorecognizeactivitiesindependentof1.INTRODUCTIONtheuser[9].Variousbiometricmodalities,suchassignature[12],Portabledevices,suchassmartphonesandpersonalvoice[13,1
7、4]andfingerprints[15]havebeenproposedfordigitalassistants(PDAs)playanimportantroleinoursecuringpersonalportabledevices.Fingerprintandeverydaylife.Thesedevicescontainincreasingamountofsignaturebasedmethodsre
