资源描述:
《外文翻译-基于emd的红外图像目标检测方法..doc》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、英文翻译系别专业班级学生姓名学号指导教师EMDBasedInfraredImageTargetDetectionMethodHeDeng&JianguoLiu&HongLiAbstractUnderthecomplicatedbackgroundofinfraredimage,thesmalltargetdetectionisavitalchallengingtaskinmodernmilitary.Inordertosolvethisproblem,anovelmethodbasedontheempiricalmodedecomposit
2、ion(EMD)isproposedinthepaper,todetectsmalltargetsundercomplicatedsea-skybackground.Thedetectionprocesscontainstwosteps:thefirststepistosuppressthesea-skybackgroundoftheinfraredimagebasedonEMD;thesecondstepistosegmentthetargetfromthebackgroundsuppressedimagethroughathreshol
3、d.Theapplicationofinfraredimageshasshownthattheperformanceofthealgorithmcandetectinfraredsmalltargetundersea-skybackgroundexactly.Comparedwithwavelettransformation,thetestingresultsbasedonEMDmethodachievetantamountresultswavelettransformation,andevenbetterinsomerespects.Th
4、esimulationsshowthatEMDmethodpresentedinthispaperappearsinstructiveforboththeoreticalandpracticalpointsofview.Keywords:Smalltargetdetection,Infraredimage,Backgroundsuppression,Threshold,EMD.1IntroductionUnderthecomplicatedbackgroundofinfraredimage,thesmalltargetdetection,i
5、dentificationandtrackingapplicationsinmodernmilitaryarevitalchallengingtasks.Ithasbeenresearchingformanyyearsoninfraredimagingsystemsandautomatictargetdetection[1].Thecomplexityofthisproblemariseswhenthetargetissmall,faintandpartiallyobscuredbysurroundingobjects,embeddedin
6、clutters.Inthesecomplexsituations,thefeaturesofboththetargetandthebackgroundaregenerallynonseparableintheoriginalimagespace.Itisdifficultforthosedetectionalgorithmstoworkintheoriginalimagespace.So,theimagehastobetransformedintoaso-calledfeaturespace,inwhichthosefeaturescan
7、beseparated.Sincetargetsandcluttershavedifferentspatialfrequencycharacteristics,aspatialfiltercouldbedesignedtosuppressanddetecttargets.BhanuandJonessummarizedalotofalgorithmsforautomatictargetdetectioninstaticinfraredimagesthatweredevelopeduptotheearly1990s[2].Thesealgori
8、thmspredominatelyutilizetraditionalimage-processingapproachesforopticalpicturesprocessing