基于數(shù)字圖像處理技術(shù)的邊緣特征提取-畢業(yè)論文外文翻譯

1、外文資料外文資料EdgeFeatureExtractionBasedonDigitalImageProcessingTechniquesI.INTRODUCTIONTheedgeisasetofthosepixelswhosegreyhavestepchangerooftopchangeitexistsbetweenobjectbackgroundobjectobjectregionregionbetweenclementclement

2、.Edgealwaysindwellsintwoneighbingareashavingdifferentgreylevel.Itistheresultofgreylevelbeingdiscontinuous.Edgedetectionisakindofmethodofimagesegmentationbasedonrangenoncontinuity.Imageedgedetectionisoneofthebasalcontents

3、intheimageprocessinganalysisalsoisakindofissueswhichareunabletoberesolvedcompletelysofar.Whenimageisacquiredthefactssuchastheprojectionmixaberrancenoiseareproduced.Thesefactsbringonimagefeaturesblurdisttionconsequentlyit

4、isverydifficulttoextractimagefeature.Meoverduetosuchfactsitisalsodifficulttodetectedge.Themethodofimageedgeoutlineacteristicsdetectionextractionhasbeenresearchhotinthedomainofimageprocessinganalysistechnique.Edgefeaturee

5、xtractionhasbeenappliedinmanyareaswidely.Thispapermainlydiscussesaboutadvantagesdisadvantagesofseveraledgedetectionoperatsappliedinthecableinsulationparametermeasurement.Indertogainmelegibleimageoutlinefirstlytheacquired

6、imageisfiltereddenoised.Intheprocessofdenoisingwavelettransfmationisused.thendifferentoperatsareappliedtodetectedgeincludingDifferentialoperatLogoperatCannyoperatBinarymphologyoperat.Finallytheedgepixelsofimageareconnect

7、edusingthemethodofbderingclosed.Thenaclearcompleteimageoutlinewillbeobtained.II.IMAGEDENOISINGAsweallknowtheactualgatheredimagescontainnoisesintheprocessoffmationtransmissionreceptionprocessing.Noisesdeteriatethequalityo

8、ftheimage.Theymakeimageblur.manyimptantfeaturesarecoveredup.Thisbringslotsofdifficultiestotheanalysis.Therefethemainpurposeistoremovenoisesoftheimageinthestageofpretreatment.Thetraditionaldenoisingmethodistheuseofalowpas

9、sbpassfiltertodenoise.Itsshtcomingisthatthesignalisblurredwhennoisesareremoved.Thereisirreconcilablecontradictionbetweenremovingnoiseedgemaintenance.Yetwaveletanalysishasbeenprovedtobeapowerfultoolfimageprocessing.Becaus

10、eWaveletdenoisingusesadifferentfrequencybpassfiltersonthesignalfiltering.Itremovesthecoefficientsofsomescaleswhichmainlyreflectthenoisefrequency.Thenthecoefficientofeveryremainingscaleisintegratedfinversetransfmsothatnoi

11、secanbesuppressedwell.Sowaveletanalysiscanbewidelyusedinmanyuseofedgeenhancementoperatfirstly.Thenwedefinethe`edgeintensityofpixelsextractthesetofedgepointsthroughsettingthreshold.Butthebderlinedetectedmayproduceinterrup

12、tionasaresultofexistingnoiseimagedark.Thusedgedetectioncontainsthefollowingtwoparts:1)Usingedgeoperatstheedgepointssetareextracted.2)Someedgepointsintheedgepointssetareremovedanumberofedgepointsarefilledintheedgepointsse

13、t.Thentheobtainedareconnectedtobealine.ThecommonusedoperatsaretheDifferentialLogCannyoperatsBinarymphologyetc.A.DifferentialoperatDifferentialoperatcanoutstgreychange.Therearesomepointswheregreychangeisbigger.thevaluecal

14、culatedinthosepointsishigherapplyingderivativeoperat.Sothesedifferentialvaluesmayberegardedasrelevant`edgeintensitygatherthepointssetoftheedgethroughsettingthresholdsfthesedifferentialvalues.Firstderivativeisthesimplestd

15、ifferentialcoefficient.Supposethattheimageisf(xy)itsoperatisthefirstderpartialderivative.Theyrepresent?f?x?f?ytherateofchangethatthegrayfisinthedirectionofxy.Yetthegrayrateofαchangeinthedirectionofaisshownintheequation(1

16、):1)?f?α=?f?xcosα?f?ysinα(Underconsecutivecircumstancesthedifferentialofthefunctionisdf=?f?xdx?f?y.Thedirectionderivativeoffunctionf(xy)hasamaximumatacertainpoint.dythedirectionofthispointisarctan[].Themaximumofdirection

17、derivativeis?f?y?f?x.Thevectwiththisdirectionmodulusiscalledasthegradientof(?f?x)2(?f?y)2thefunctionf，thatis.Sothegradientmodulusoperatisdesigned?f(xy)=(?f?x?f?x)intheequation(2).(2)G[f(xy)](?f?x)2(?f?y)2Fthedigitalimage

18、thegradienttemplateoperatisdesignedas:(3)DifferentialoperatmostlyincludesRobotsoperatSobeloperat.(1)RobertsoperatRobotsoperatisakindofthemostsimpleoperatwhichmakesuseofpartialdifferenceoperattolookfedge.Itseffectisthebes