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(Contentbasedmultimediaretrieval)ContentbasedmultimediaretrievalBasedontherelatedconceptsandcharacteristicsofcontent-basedmultimediaretrievalareintroduced,theanalysismethodbasedoncontent,inthecompresseddomainbasedonMPEGaudiosignal,toachievereal-timemultimedi
2、aanalysisandretrievalpurposes.Thealgorithmconsistsofthreesteps:first,usingcompresseddomainfeaturesforaudiosignalsegmentation,andthenapplythehierarchicalmethodtosplitupaudioclipsintoroughmusic,voiceandotherthreebasiccategories;thespeakeridentityisimportantretrievalcues
3、inthespeechsignal,thenusingthehiddenMarkovchaintoachievethetextindependentspeakerrecognition,andtheidentityofthespeechsignalandthecorrespondingvideoannotationwithrecognizedwords.Keywordsaudioretrieval,conceptmultimedia,contentbasedretrieval,compresseddomain,hiddenMark
4、ovchain,speakerrecognition,multimediaretrievalIntroductionWiththedevelopmentofcomputerapplicationtechnologyandtheincreaseofInternetspeed,multimediainformationsuchastext,audioandvideocanbeaccessedcontinuously.Thus,themainproblemsfacedbycomputerusersindealingwithinforma
5、tionhavechangedfromthelackofearlyinformationtotherapidandrationalretrievalofneededinformationfromvastamountsofinformation.Thus,fromthebeginningof90s,content-basedimageretrievalbecameoneofthefocusesinmultimediaresearchfield[1][2].Incontentbasedimageretrieval(video),col
6、or,texture,shapeandmotionofthevisualfeaturesareextractedtodescribeimage(video)semanticcontentcontains,inordertoachievetheimage(video)queryanddatamanagement.PrinciplesandcharacteristicsofcontentbasedmultimediaretrievalMultimediaretrievalisacontent-basedfeatureretrieval
7、(CBR:content-based,retrieval).Theso-calledcontentbasedretrievalistoretrievethecontentofthemediaobjectsandthesemanticenvironment,suchascolor,texture,shape,videoshot,scene,shotmotion,voicetone,loudness,timbreetc..Contentretrievalbreaksthroughthetextretrievaltechnologyba
8、sedonthelimitationofthetraditionalbaseddirectlyontheimage,videoandaudiocontentanalysis,featureextractionandsemanticfeatures,