基于深度圖像的人體關(guān)節(jié)點(diǎn)定位的方法研究.pdf

1、聲明本學(xué)位論文是我在導(dǎo)師的指導(dǎo)下取得的研究成果，盡我所知，在本學(xué)位論文中，除了加以標(biāo)注和致謝的部分外，不包含其他人已經(jīng)發(fā)表或公布過(guò)的研究成果，也不包含我為獲得任何教育機(jī)構(gòu)的學(xué)位或?qū)W歷而使用過(guò)的材料。與我一同工作的同事對(duì)本學(xué)位論文做出的貢獻(xiàn)均已在論文中作了明確的說(shuō)明。研究生簽名：呂弗掃I牛年弓月萬(wàn)日學(xué)位論文使用授權(quán)聲明南京理工大學(xué)有權(quán)保存本學(xué)位論文的電子和紙質(zhì)文檔，可以借閱或上網(wǎng)公布本學(xué)位論文的部分或全部?jī)?nèi)容，可以向有關(guān)部門(mén)或機(jī)構(gòu)送交并授

2、權(quán)其保存、借閱或上網(wǎng)公布本學(xué)位論文的部分或全部?jī)?nèi)容。對(duì)于保密論文，按保密的有關(guān)規(guī)定和程序處理。研究生簽名：邑事厶Ⅳ年弓月巧日碩士學(xué)位論文竺竺!——————————————————————————一AbstractH啪a11P。seRec。gniti。nisahotsp。t。fcurrentc。mputerVisi。nstudyWitllmegrowmofcomputerso胍Ireandhardwarefields，humanPoser

3、ecognitiontechnologYisinc陀勰1nglyadhibitedingameaIldmonitoringareasDuringtheprocessofh啪anpos㈨aly818’po∞斌thi曲1yv撕e二，a11dtheyarehiddenunderc1。thesA11thesefact。rsenhance‘hedi街cultyofh砌aIlmotion眥alysis，whichbringschallengesan

4、dopportunitiestOthe鏟。礎(chǔ)oth眥觚Doserecognitionstudy‘Thebinhanddevel。pment。fdepthimagesbringshumanp。serec。gniti。ntechnologYfrom1aboratorytopeopleseveryday1iifnef0I咖maatgie。snc，awptullircehdbisyndotepthafrecacmteedrabdyofanoct

5、t。crosntlaikinelcoorinformation，butonlydistance1n士?？tlon，w工ucn1s‘1u?！痙上塢uujsunshine，Shad。wa11detcThis，t0s。meextend，makesiteasiert。rec。gnizeandforecaStThispapermakescontributionsasfollows：Thefirstchapterbfieflyintroduces

6、thedevelopmentofhumaJlposerecognitiontechnologyandthetheoryofdepth1magmgtechnology，analyzessometypicalhumanposerecognitionalgorimmsand∞mp嬲‘：lech麟teriSticsb咖eendepthimagesandcolorfulimagesThesecondchaptertalksabom’hemetho

7、d。fm狄ingsegIllentati。nbetweenbackgr。undandpe。pleusingRANsAcalgoritmtofitthemodelofceflingandfloorfromcollecteddepthimages，狃dutilizing∞nto哪follo謝ngteclllliqueaJlddepthinformationofobjectstoextrac‘humaIlboayareasfrombackgr

8、oundimagesThefollowingchapterdiscussesusingrandomforestQlasslfiertoconductcategorizationforeachpixelinhlmanbodyarea，andafterwardsSOrtmemintodlfierentbodycomponentsInp砒icul甄themethodofextractingcharacteristicsisbythehelpo

9、feachDixelsoffseta1191einformationtogainlocaldepthgradientcharacterThischaractercanensurematmepixelwontchangeinthemoVingandrotatingconditionThefourthchapterstudiesthewaytoapplymeanshiftscorerankingmethod，whichaimstofindt

10、hedensestpixelpointinprobabilitytostandforthepositionofjoin‘a(chǎn)mongthesame’typebodycomponentpixelsOnthisbasis，thispaperusesprioriknowledgeaboutmotlonconstraIntstooptimizetheproposalsofjoints’positionsThenextchapterreponsan

11、d猢ly聆鶴m餾o：aualltitativea11dqualitativetestsofthesystemanddrawsconclusionsbasedonmosedataandcomparesthispaperwithotheralgorithmThesixthchapters咖nanzesthesystemsperfornlaIlceandmakesavisionofthisfield。futurityTheaVeragetim