視頻序列中高精度目標(biāo)檢測與融合.pdf

1、南京理工大學(xué)碩士學(xué)位論文視頻序列中高精度目標(biāo)檢測與融合姓名：朱冰申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：婁震201103AbstraCtMoviIlgobjectdetectingisthe‰d鋤entaltaSkinvide0processing，whjchh弱playedaveryimportantroleinVideocoding，video吼玎Veillallce，multimediaiIldices，robotnavi

2、g撕onandotherfileds；whileimagefIlsiontechniqueh觴been麗delyusedi11inlage礎(chǔ)ting，TVandfilmproductionConsideringthedemandt0addaVirtualobjectinVideosc饑e，mispapermakessomediscussionsa11dresearchs0nmoVingobjectdetecting趾dimage缸sio

3、nteclmiqueThispaperf0cllsesonbackgroundSub仃actionalgorim【nandsynmetricaldi婦eIlcingal鰣thInThroughaIlalysisingandcomp撕ngoftherelatedbackgromldrecons饑ctionmetllods，tllispaperchoosesthe嗽Itistichisto莎ammethodt0recons饑lctionme

4、initialbackgr01lnd，tllcllcombinesbackgro岫dsub訂actionalgorithma11dsyll衄嘶caldi岱。rencingalgori讓lmt0detectmoVingobjecttog甜1％whichcanoV即comethedisadVantagesofthatbackgrolllldsub仃actionmaymisstlleobjeCtwrhoseValueiscloset0back

5、grollnd’s鋤dsymmetricaldi鼢cingmayproduce鋤ptyholeinacem血eXtentInordert0adaptmechaIlgeofbackgroundinthescene，areal—timeaIldap耐odupdatingm砒odsarebo廿lproposedtoupdatemebackgroundaCcordingtotheobjectm勰kWheIlbackgro吼dsubtra嘶ona

6、lgorithmands弘吼etricaldi儂舭galgori伽[Ilbotllc吼’tdetectpartofmeobject，alispaperproposesasecondtimedetectionme讓10dt0getthecomplcteobjectaccordingtomerelativityofconsecutiVe丘鋤eiIlⅥdeose(1uenceTbsuppresstheimpactofshadow，usenom

7、alizedRGBcolormodelandLBPt旬m”eoperatortodetectshadow，缸a11ycX仃actaccu期tcmovingobjeCtthrou曲morphologyprocess證g鋤dcormecteddomainanalysisT0mal【ethevirtllalobiectlookmorenanlraIandn0obvioussi弘sofcllt趴dpasteinnlesccnedu血gtllef

8、Ilsionofvideohage，estimatethea_lphaValueofeach0fthem0Vingobjectpixels，a11dmenproduceanewvide0sequencewhichisaddcdaVimlalobjectfinallVKeyword：M0viI培dbject，baCk伊0undrecons仃uction，backgroundsllb昀ction，s舯metricaldif|衙eIlCing