资源描述：
《基于内容的多媒体检索（content-based multimedia retrieval）》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。
1、基于内容的多媒体检索（Content-basedmultimediaretrieval）Content-basedmultimediaretrievalOfcontentbasedmultimediaretrievalrelevantintroducestheconcept,characteristics,basedonthecontentanalysismethodisputforward,inthecompresseddomaindirectlyanalyzetheMPEGaudiosignal,toachievemultimediareal-timeanalysi
2、sretrieval.Algorithmisdividedintothreesteps:firstuseofcompresseddomaincharacteristicsofaudiosignalissplit,andthenlayeredmethodisappliedtothesegmentationofaudioclipsintomusic,voiceandotherthreebasiccategories;Duetoidentitythespeaker'swordsisimportantinspeechsignalretrievalclues,finallyusi
3、nghiddenmarkovchaintorealizethemanwhohasnothingtodowiththetextwordrecognition,andidentifiesoutidentityofspeechsignalanditscorrespondingvideoannotation.Keywordsaudioretrievalconceptmultimediacontent-basedretrievalcompressiondomainhiddenmarkovchainspeakeridentifiesmultimediaretrievalTheint
4、roductionWiththedevelopmentofcomputerapplicationtechnologyandtheimprovementofInternetspeed,multimediainformationsuchastext,audioandvideocanbeaccessedbyusers.Inthisway,themajorproblemsthatcomputerusersfaceinprocessinginformationhavebeentransformedfromtheearlyinformationdeficiencytotherapi
5、dandreasonableretrievalofinformationneededfrommassinformation.Therefore,fromtheearly1990s,thecontent-basedimage(video)retrievalbecameoneofthehotspotsinthemultimediafield.[1][2].Intheimage(video)retrievalbasedoncontent,color,texture,shape,andmotioncanbeextractedvisualcharacteristics,sucha
6、scharacterizationoftheimage(video)thesemanticscontainedinthecontent,soastorealizetheimage(video)dataqueryandmanagement.TheprincipleandcharacteristicsofmultimediaretrievalbasedoncontentMultimediaretrievalisacontentbasedretrieval(CBR:content-basedretrieval).So-calledcontent-basedretrievali
7、sthemediaobjectandthecontentofthecontextsemanticretrievalenvironment,likethecolor,textureandshapeoftheimage,themovementofthevideocamera,scene,lens,pitch,loudness,toneofvoice,etc.Content-basedretrievaltobreakthroughthelimitationoftraditionaltext-basedretrievaltechnology,di
