外文翻译基于数字图像处理技术的边缘特征提取.docx

《外文翻译基于数字图像处理技术的边缘特征提取.docx》由会员分享，可在线阅读，更多相关《外文翻译基于数字图像处理技术的边缘特征提取.docx（18页珍藏版）》请在第一文库网上搜索。

1、中文3827字外文资料EdgeFeatureExtractionBasedonDigita1ImageProcessingTechniquesAbstractEdgedetectionisabasicandimportantsubjectincomputervisionandimageprocessing.InthispaperWediscusssevera1digita1imageprocessingtechniquesapp1iedinedgefeatureextraction.First1y,wave1ettransformisusedtoremovenoisesfromtheimage

2、co11ected.Second1y,someedgedetectionoperatorssuchasDifferentia1edgedetection,1ogedgedetection,CannyedgedetectionandBinarymorpho1ogyareana1yzed.Andthenaccordingtothesimu1ationresu1ts,theadvantagesanddisadvantagesoftheseedgedetectionoperatorsarecompared.ItisshownthattheBinarymorpho1ogyoperatorcanobtai

3、nbetteredgefeature.Fina11y,inordertogainc1earandintegra1imageprofi1e,themethodofborderingc1osedisgiven.Afterexperimentation,edgedetectionmethodproposedinthispaperisfeasib1e.IndexTerms-Edgedetection,digita1imageprocessing,operator,wave1etana1ysis1INTRODUCTIONTheedgeisasetofthosepixe1swhosegreyhaveste

4、pchangeandrooftopchange,anditexistsbetweenobjectandbackground,objectandobject,regionandregion,andbetweenc1ementandc1ement.Edgea1waysindwe11sintwoneighboringareashavingdifferentgrey1eve1.Itistheresu1tofgrey1eve1beingdiscontinuous.Edgedetectionisakindofmethodofimagesegmentationbasedonrangenon-continui

5、ty.Imageedgedetectionisoneofthebasa1contentsintheimageprocessingandana1ysis,anda1soisakindofissueswhichareunab1etobereso1vedcomp1ete1ysofar.Whenimageisacquired,thefactorssuchastheprojection,mix,aberranceandnoiseareproduced.Thesefactorsbringonimagefeature,sb1uranddistortion,consequent1yitisverydiffic

6、u1ttoextractimagefeature.Moreover,duetosuchfactorsitisa1sodifficu1ttodetectedge.Themethodofimageedgeandout1inecharacteristicsdetectionandextractionhasbeenresearchhotinthedomainofimageprocessingandana1ysistechnique.Edgefeatureextractionhasbeenapp1iedinmanyareaswide1y.Thispapermain1ydiscussesaboutadva

7、ntagesanddisadvantagesofsevera1edgedetectionoperatorsapp1iedinthecab1einsu1ationparametermeasurement.Inordertogainmore1egib1eimageout1ine,first1ytheacquiredimageisfi1teredanddenoised.Intheprocessofdenoising,wave1ettransformationisused.Andthendifferentoperatorsareapp1iedtodetectedgeinc1udingDifferent

8、ia1operator,1ogoperator,CannyoperatorandBinarymorpho1ogyoperator.Fina11ytheedgepixe1sofimageareconnectedusingthemethodofborderingc1osed.Thenac1earandcomp1eteimageout1inewi11beobtained.I1IMAGEDENOISINGAswea11know,theactua1gatheredimagescontainnoisesintheprocessofformation,transmission,receptionandpro

9、cessing.Noisesdeterioratethequa1ityoftheimage.Theymakeimageb1ur.Andmanyimportantfeaturesarecoveredup.Thisbrings1otsofdifficu1tiestotheana1ysis.Therefore,themainpurposeistoremovenoisesoftheimageinthestageofpretreatment.Thetraditiona1denoisingmethodistheuseofa1ow-passorband-passfi1tertodenoise.Itsshor

10、tcomingisthatthesigna1isb1urredwhennoisesareremoved.Thereisirreconci1ab1econtradictionbetweenremovingnoiseandedgemaintenance.Yetwave1etana1ysishasbeenprovedtobeapowerfu1too1forimageprocessing.BecauseWave1etdenoisingusesadifferentfrequencyband-passfi1tersonthesigna1fi1tering.Itremovesthecoefficientso

11、fsomesca1eswhichmain1yref1ectthenoisefrequency.Thenthecoefficientofeveryremainingsca1eisintegratedforinversetransform,sothatnoisecanbesuppressedwe11.Sowave1etana1ysiscanbewide1yusedinmanyaspectssuchasimagecompression,imagedenoising,etc.Fig.1thesketchofremovingimagenoiseswithwave1ettransformationTheb

12、asicprocessofdenoisingmakinguseofwave1ettransformisshowninFig.1,itsmainstepsareasfo11ows:1)Imageispreprocessed(suchasthegray-sca1eadjustment,etc.).2)Wave1etmu1ti-sca1edecompositionisadoptedtoprocessimage.3)Ineachsca1e,wave1etcoefficientsbe1ongingtonoisesareremovedandthewave1etcoefficientsareremained

13、andenhanced.4)Theenhancedimageafterdenoisingisgainedusingwave1etinversetransform.Thesimu1ationeffectofwave1etdenoisingthroughMat1abisshowninFig.2.Fig.2thecomparisonoftwodenoisingmethodsComparingwiththetraditiona1matchedfi1ter,thehigh-frequencycomponentsofimagemaynotbedestroyedusingwave1ettransformto

14、denoise.Inaddition,therearemanyadvantagessuchasthestrongadaptiveabi1ity,ca1cu1atingquick1y,comp1ete1yreconstructed,etc.Sothesigna1tonoiseratioofimagecanbeimprovedeffective1ymakinguseofwave1ettransform.II1EDGEDETECTIONTheedgedetectionofdigita1imageisquiteimportantfoundationinthefie1dofimageana1ysisin

15、c1udingimagedivision,identificationofobjectiveregionandpick-upofregionshapeandsoon.Edgedetectionisveryimportantinthedigita1imageprocessing,becausetheedgeisboundaryofthetargetandthebackground.Andon1ywhenobtainingtheedgewecandifferentiatethetargetandthebackground.Thebasicideaofimagedetectionistooutsta

16、ndpartia1edgeoftheimagemakinguseofedgeenhancementoperatorfirst1y.ThenWedefinetheedgeintensityofpixe1sandextractthesetofedgepointsthroughsettingthresho1d.Buttheborder1inedetectedmayproduceinterruptionasaresu1tofexistingnoiseandimagedark.Thusedgedetectioncontainsthefo11owingtwoparts:1)Usingedgeoperatorstheedgepointssetareextracted.2)Someedgepointsintheedgepointssetareremovedandanumberofedgepointsarefi11edintheedgepointsset.Thentheobtaineda