基于貝葉斯推理的視頻語(yǔ)義自動(dòng)標(biāo)注.pdf

1、北京交通大學(xué)碩士學(xué)位論文基于貝葉斯推理的視頻語(yǔ)義自動(dòng)標(biāo)注姓名：王潔申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)技術(shù)與應(yīng)用指導(dǎo)教師：王方石20061201Ib立鑾道太堂亟堂焦i金塞△旦5至堡△￡!ABSTRACTWimthedevelopmeIltofcomputernetwork姐dmultimediatecl】Ilology，Vidcodatagrowwitllmeu印reced鋤ted印eedsHowcverVideoillfb肌ationm觚a

2、gcIncllttcchnoIogyonlycanprovidenon一1inear、Ilon—intelligentoperationforcustomerwtlichc鋤otsatisfythenecdofaccess觚dmaIlag鋤emofV勰tandrichVideomforInation，m私siveiIlfbrrnationc鋤notbereusednbecomesaprobleIIltobes01vedhowtealiz

3、cautomaticors鋤i肌tomaticVidcoinfomationm鋤agemcnt鋤dprovidccoIⅣeIlieIltiIlfo塒衄虹onmanag鋤entsystemUnder鋤chback目ound，ComentBascdvideoAnalySis鋤dRc仃i吖al(CB、AR)tecllIlologyisproposcdtoanalyze、orgamzeaIldmaIlagctllevaStvide0data’C

4、BⅥ～RisaefI＆tive∞1utionformeaboveconnictandbecomeshotpoimintlleresearchfieldofmultimediatcctlIlologynow111elow1eVelvisualfeatIl】嗚sofVid∞arestaticfbatureincludingc010r，textll】re，andshape，etcButthedescription孤Idunderstalldi

5、ngof“d∞forpcoplearcm砌y∞s鋤anticleVcl，complexi咄fuzzinessandsubjectivityofvideosemamiccomemlead”semanticg印”be帆e鋤low一1evelfcaturc觚dhi曲一levelfbatIlrc，moreovcrtllisprobl鋤haveIl’tbecll∞lVedverywellTheref研ehowt0getVideoselIl礬tic

6、鋤otationautomaticallybecomesavitaltopichlmispap％t11eexisting“dcos鋤鋤tica衄otationtechn0109yisde印ly觚alyzedaIldamethodisprclposedto秈otatenon—fixedlen毋h鋤otationfor、，ideosFirst，Bayesi柚net、Ⅳorkisestablishedtoexpm昭thecO—occ嘶ngrc

7、latio璐11ipofs鋤鋤ticconc印tsThen，NaWeBayesi鋤cl夠smerisusedtoobtailltlleseIllallticcaIldidatesetofvideoFinally，Bayesiallhlf醯朗ceisIIscdt0gctmefinal釉otationset矗∞mthes鋤a11ticcalldidatesctThee】【p舒melltalresunindicatestheproposedm