基于顏色特征的圖像檢索方法的研究及系統(tǒng)實現(xiàn)

1、曲阜師范大學(xué)碩士學(xué)位論文基于顏色特征的圖像檢索方法的研究及系統(tǒng)實現(xiàn)姓名：程潔申請學(xué)位級別：碩士專業(yè)：計算機應(yīng)用技術(shù)指導(dǎo)教師：趙景秀20090401皋于顏色特征的圖像榆索方法的研究及系統(tǒng)實現(xiàn)AbstractWiththerapiddevelopmentofthemultimediatechnologyandthenetworktechnologyimagedatasourcescontinuetoexpand，andtheworld’Sd

2、igitalimagecapacityisgrowingatanalarmingrateThesedigitalimagescontainagreatdealofusefulinformation，inordertofindthecontentofuserrequirementsfromthemassivedatabasequicklyandaccuratelyContentbasedimageretrieval(CBIR)techno

3、logyhasbeenwidelyconcernedandbecomeahotspotofresearchContent—basedimageretrievalmakestheuseofcolortexture，shapeandotherunderlyingcharacteristicsandsemanticofotherhigh—levelcharacteristicstosearchimageSincethebackgroundof

4、theimageusuallyoccupythelargerpartoftheimage，impactsthefeaturesofthemainobjectiveofimageintheprocessoffeatureextraction，imagesdescribedbasedonglobalfeaturesarenotwellreflecttheimageofthesemanticfeatureClassicalcolorbased

5、histogramiseasytocompute，insensitivetotranslation，rotationandscale，andquiteresistanttonoisesaswellButitcan’treflectspatialinformationofthecoloLpronetofalsehitswhendistinguishingimagesinlargedatabasewithsimilarcolorcompos

6、itionbutdifferentspatialdistributionInordertosolvethisproblem，afterreferencingtoalargenumberofthelatestresearchresultsandconductindepthexplorationexperiments，Thisarticleoutlinesthebackgroundandsignificanceofimageretrieva

7、l，researchsituationandapplicationareas，reviewsthesystemstructureoftheimageretrieval，thetechnicalclassificationoftheretrievalmethod，similaritymeasurementmethodsandsystemperformanceevaluation，focusesonintroducingtheimagere

8、trievalbaseoncharacteristicsofcoloLandwepresentanimageretrievalmethodbasedoncolorfeatureandshapefeatureItcanuseOtsumethodtodeterminetheappropriatethreshold，getthebackgroundandthetargetimagebyimagesegmentationAsforcolorba

9、sedimageretrieval，thepaperselectHSVspace，atthesametimethetransformationbetweenRGBspaceandHSVspaceispresentedInorderteducethedimensionofcolorfeature，quantificationofthecolorspaceteducethedimensionImageblockfirstlymakingth

10、euseofanimprovedfansegmentationstrategyandthenfeatureextractcombiningcolorhistogram，andusethesefeaturestoimageretrievalThismethodnotonlycanruleouttheinterferenceofbackgroundpixels，butalsopositionthepixelofthecolorpositio

11、nofthetargetimageExperimentalresultsshowthatthisimprovedapproachisbetterthanthesimpleuseofcolorhistograminanaccuratesurveyKeywords：content—basedimageretrieval；colorhistogram；imagesegmentation；featureextraction；Eigenvecto