基于毫米波雷达的身份识别技术研究.pdfVIP

浙江大学硕士学位论文ABSTRACT

Abstract

Identityrecognitiontechnologyhasawiderangeofapplicationsindailylife,suchascom-

puterunlocking,mobilephonepaymentandsystemlogintoverifyyouridentityallthetime.Traditional

identificationmethodsmainlyincludefingerprintrecognition,facerecognition,passwordrecog-

nitionandsoon.Fingerprintidentificationrequiresdirectcontactbetweenhumanfingersandthe

sensor,whichisapotentialhealthhazard.Facerecognitionneedstocollecteachperson’sfacein-

formation,therearehiddenrisksofprivacydisclosure;Passwordrecognitionrequirespeopleto

remembertheirownpasswordsratherthanusingthebody’sowncharacteristics,thereisariskof

forgetting.

ThisthesisputsforwardaseriesofbasedonFMCWMillimeterWaveRadaridentityrecog-

nitiontechnology.Byanalyzingthehumanradarechosignalandusingthecorrespondingdeep

learningclassifiertoidentifytheindividualidentity.Thetechnologyallowsforcontactless,bio-

metricandprivacy-freeidentification.

Theworkincludes:

1.ThisthesisstudiesthetheoreticalbasisofFMCWmillimeterwaveradar,andexplainshowto

obtainthedistance,velocity,Angleandotherinformationoftheobjectthroughtheoriginal

echosignalofFMCWmillimeterwaveradar.Theradartargetdetectionwasstudied,andthe

backgroundsubtractionmethodandConstantFalse-AlarmRate(CFAR)werecompared.

2.Thisthesisstudiesandproposesanidentityrecognitiontechnologybasedonsinglepalm/head

radarechosignal,whichincludestargetdetection,staticdetection,invivodetection,data

preprocessing,modelreasoning,multi-Chirpjointjudgmentandothersteps.Thethesispresents

thepreprocessingmethodofradarechodatanormalizationofPNAand