视频图像文字检测分析-text detection and analysis of video images.docxVIP

  • 5
  • 0
  • 约3.41万字
  • 约 55页
  • 2018-08-01 发布于上海
  • 举报

视频图像文字检测分析-text detection and analysis of video images.docx

视频图像文字检测分析-text detection and analysis of video images

AbstractAlongwiththeadvanceofdigitalprocess,Multimediaandcommunicationsindustryhavefoundwideapplicationinourdailylife.Digitalvideosandimageshavebeenasthemainmedia,whichisomnipresentnowadaysandincreaseconstantly.Tomeettheincreasinglydiverserangeofrequirementsofthemassesofthepeople,thetechnologyofdigitalvideoprocessinghasmadesubstantialprogress.Therearesomanydigitalvideosineverydaylife,includingdigitalcinema,networkvideoimage,andmedicalimageandsupervisingvideos.Processingagreatdealofthosemultimediadatahasdrawnworldwideattentionandwecanseegreatpotentialinthis.Recently,studyfortextdetectionfromvideoimageshasbeenahotspotintheareaofimageprocessingandanalysis.Thisthesis,focusingontextdetectionandsegmentationfromvideoimages,makearemarkableresearch.Themaincontentofthisthesisinclude3aspectsasbelow:Inthisthesis,authorfirstlyanalysisandcomparessomecommonlyusedtextdetectionandsegmentation,andpointsoutsimilaritiesanddifferencesbetweenthem.Researchingontextdetectionandlocalization,thisthesispresentsanovelandeffectivemethodtodetectandlocatetextlinefromvideoimagehavingcomplexbackground.Byanalyzingthecomplexityofbackground,videoimageshasbeenclassifiedanddescribedasbelow:simplebackground,normalbackgroundandcomplexbackground.Differentbackgroundshavedifferentadaptivemethod,whichcanassureaccuracyaswellasincreaseefficiency.Especially,forthecomplexbackgroundimages,wemakethoroughanalysisandresearch,andadoptseveralsuccessionvideoimagestoremovefalsealert.Increasetheefficiencyandprecisionofthenovelmethod.Researchingontextsegmentation,thisthesispresentasystematictextsegmentationalgorithmbasedoncommonlyusedtextsegmentation,whichincludesanalyzehorizontalprojectionandverticalprojection,judgethepolarofthetextarea,multiframeverificationandstrokesegmentation.Fortheabovealgorithm,thethesishasgatheredalargenumberofvideoimagesandmadealotoftesting.Testingresultsshowthattheproposedmethodhasachievedgoodresultsintermsofefficiencyandeffectoftheperformance.Keywords:TextdetectionConditionmorphologyMultiframeve

您可能关注的文档

文档评论(0)

1亿VIP精品文档

相关文档