文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > 自然科学 > 物理 > 图像处理Reaction–diffusion network for geometric multiscale high speed image processing
侵权投诉

图像处理Reaction–diffusion network for geometric multiscale high speed image processing

图像处理经典论文 EI SCI

ImageandVisionComputing28(2010)

图像处理Reaction–diffusion network for geometric multiscale high speed image processing

914–926

ContentslistsavailableatScienceDirect

ImageandVisionComputing

journalhomepage:doc.guandang.net

图像处理Reaction–diffusion network for geometric multiscale high speed image processing

s

Reaction–diffusionnetworkforgeometricmultiscalehighspeedimageprocessing

S.Binczaka,*,T.Sliwab,S.Jacquira,J.M.Bilbaulta

ab

LE2ICNRSUMR5158,AiledesSciencesdel’Ingénieur,UniversitédeBourgogne,BP47870,21078DijonCedex,FranceLE2ICNRSUMR5158,UniversitédeBourgogne,Routedesplainesdel’Yonne,BP16,89010Auxerre,France

articleinfoabstract

Intheframeworkofheavymid-levelprocessingforhighspeedimaging,anonlinearbi-dimensionalnet-workisproposed,allowingtheimplementationofactivecurvealgorithms.Usuallythisef cienttypeofalgorithmisprohibitiveforreal-timeimageprocessingduetoitscalculuschargeandtheinadequatestructurefortheuseofserialorparallelarchitectures.Anotherkindofimplementationphilosophyispro-posedhere,byconsideringtheactivecurvegeneratedbyapropagationphenomenoninspiredfrombio-logicalmodeling.Aprogrammablenonlinearreaction–diffusionsystemisproposedunderfrontcontrolandtechnologicalconstraints.Geometricmultiscaleprocessingispresentedandthisopensadiscussionaboutelectronicimplementation.

Ó2009ElsevierB.V.Allrightsreserved.

Articlehistory:

Received12July2007

Receivedinrevisedform11October2009Accepted19November2009

Keywords:

Imageanalysis

Multiscalegeometry

Nonlinearsignalprocessing

1.Introduction

Thebackgroundframeworkofthisworkconsistsofthedevelopmentofprocessingsystemsandtheirelectronicimple-mentationforhighspeedimageintelligentsensors.Nowadays,manyimageprocessingtechniques,particularlymostlinearorprobabilisticones,areincludedinreal-timesystems,butmanyfamiliesofmid-leveltechniquesstillremainunusableforsuchtimeprocessingrequirements.

Forhighspeedimagingapplications,onlysomefamiliesofsimplelow-leveltechniquescanbeused,othertechniquesbeingappliedonrecordeddata.Forhighspeedimagingtreatments,theapproachofarti cialretinasconsistinginlocatingprocessingatthepixelstagegivessomegoodresults[1–3].However,asigni -cantdisadvantagelieswiththisapproach:manytreatmentseitherremainverydif culttoadapttoretinaarchitecturesorimplydevelopmentofspeci carchitectures.Toavoidthisdisadvantage,agenerictypeofimageprocessingisproposed.

Mathematicalmorphology[4,5]hasbeenspeci callydevelopedtomeetimageprocessingexpectations.Ithelpstosolveamajorpartofimageprocessingproblemsandiswidelyused.Highspeedelectronicimplementationsofbasicmorphologicaloperatorsasdilationshavealreadybeenstudied,forinstancewiththeuseofanonlinearelectronicnetwork[6].However,themostimportantfamiliesofmathematicalmorphologytreatmentsarebasedontwoothermainheavyoperatorswhicharegeodesicreconstructionandwatershedalgorithm.

*Correspondingauthor.Tel.:+33380396848;fax:+33380395910.E-mailaddress:stbinc@u-bourgogne.fr(S.Binczak).0262-8856/$-seefrontmatterÓ2009ElsevierB.V.Allrightsreserved.doi:10.1016/j.imavis.2009.11.008

Differentarchitectureshavebeenbuiltinordertoimplementthesemid-leveltreatments[7–9],mostofthemusingnumericaltechnologies,abletodoreal-timeimageprocessingforusualspeeds,butwithatechnologicalgappreventinghighspeedimag-ingapplications.

Inthiscontext,apossiblegeneralizationofthegeodesicdilationisexploredbyconstrainingitsconstructionwithcompatibilityregardingasecondconnexitynotionlinkedtotheregularityoftheshapeoftheobjects.Itisnotexpressedalgebraically,butthroughamoreintuitiveapproach.Infact,geodesicdilationcanbeinterpretedasatopologicalwavepropagationstartingfromamarkerwhoseevolutionisde nedbythestructureoftheimage.PartialDifferentialEquations(PDE)areaveryconvenienttooltogeneratethispropagationwherefeaturesoftheimageareincludedasintrinsicparametersoftheequations.Furthermore,someofthereaction–diffusionPDEincludeanotionofregularityfortheprop-agationphenomenon.Forsomeofthem,itisevenpossibletoadjustparametersde ningthecapacityofthewavetotravelthroughstraightzones,holes,breaksandsoon.Therefore,itde nesakindofgeodesic-activecurveorregion.

Anotheraspectisveryimportantforthenotionofobjects’reg-ularity:thescalefromwhichtheobjectisobserved.Butcarryingoutmultiscaleanalysisthroughsuccessivemultiplewavepropaga-tioniscontrarytothehighspeedimagingtechnologicalgap.Inor-dertoreducetreatmenttimes,theidealcasewouldbetoachieveamultiscaleanalysiswithonlyaone-passwavepropagation.Infact,aphenomenonproducingpropagationofwavefrontswithincreas-ingregularityalongtheheightofthetransitionpartisstudied.AsystematicwayofexploringdynamicalsystemsdescribedbyPDEhasbeenobtainedbydevelopingelectroniccircuits,whichareveryef cientforreal-timeanalysis.Thesetechniquesleadto

第1页

TOP相关主题

你可能喜欢

热门文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email