基于圖分割模型的鏡頭分割和視頻廣告檢測(cè)的研究.pdf

1、濟(jì)南大學(xué)碩士學(xué)位論文基于圖分割模型的鏡頭分割和視頻廣告檢測(cè)的研究姓名：譚麗娜申請(qǐng)學(xué)位級(jí)別：碩士專(zhuān)業(yè)：模式識(shí)別與智能系統(tǒng)指導(dǎo)教師：李金屏20090520濟(jì)南大學(xué)碩士學(xué)位論文ABSTRACTWiththedevelopmentofmultimediacompressiontechnologyandcomputernetwork，therearemoreandmoremultimediainformationhowtogettherighti

2、nformationwewantexpeditiouslyandpreciselyfromsuchenormousmultimediainformationbecomeahotresearchgraduallyShotboundarydetectionasabasisforthevideetrievalhasagreatsignificanceAtpresentshotboundarydetectionismostlybasedonsi

3、mpledifferenceofthetwoadjacentvideoframesItiseasilydisturbedbynoiseThroughtheanalysisofshotboundarythemethodofgraphpartitionmodelisusedforshotboundarydetectionTVadvertisementretrievalisbecomin，gan’importantissuetodecreas

4、etheillegaladvertisementSOthetechnologyofshotboundarydetectionisappliedtoTVjt#t714att’’’advertisementretrievalThetechniquesofShotboundarydetectionconsistofthreecoreelements，therepresentationofvisualcontent，theevaluationo

5、fvisualcontentcontinuityandtheclassificationofcontinuityvaluesFirstofall，theHSVcolorhistogramisusedtepresentvisualcontent；Thengraphpartitionmodelisusedtoconstructvideostreamingsignal，thethresholdmethodisusedtoextractthet

6、roughshapedregionsasthecandidates；FinallythecutandgradualtransitionwithoutmonochromeframesareextractedbytheneuralnetworksThegradualtransitionregions、析吐lmonochromeframescouldbeextractedbythecharacterofthemonochromeframesT

7、hedetectionofcutshasaprecisionrateof9781％，thegraduationtransition謝mmonochromeframeshas9231％，othergraduationtransitionhas8889％，inall103programsInthebasisofpreliminaryworkthetechnologyofshotboundarydetectionisappliedtoTVad

8、vertisementretrievalAsweknowadvertisementhastwobasicfeatures：frequentshotchangesandapparentbrandinformationTVadvertisementisextractedbythehigherfrequencyoftheshotboundaryAsforthoseadvertisements、杭thlowshotchangefrequenci

9、es，thetextdetectionmethodisemployedtoextractthebrandinformationavideoadvertisementretrievalsystemisdesignedandrealizedThedetectionforvideoadvertisementhasaprecisionrateof8111％inthreevideoprogramsThedetectionofthecuthasab