2024年5月19日发(作者：斐讯路由器设置登录入口)

ACoherentRadonTransformforSmallTarget

Detection

JavierCarretero-Moya,JavierGismero-Menoyo,AlbertoAsensio-López,ÁlvaroBlanco-del-Campo

DepartamentodeSeñsidadPolitécnicadeMadrid

CiudadUniversitaria,s/.

Email:{jcarretero,javier,vera,alvaro}@

Abstract—TheRadontransformcanbeusedtoimplement

gridbased“track-before-detect”roachcan

bedescribedasarangealignmentstepfollowedbynon-coherent

paper,amodificationoftheconventional

directRadontransformisproposedtocoherentlyintegrate

way,thenoiseandclutter

contributionstoeachprojectionanglecanbeminimizedand

theoverallperformanceofthesystemwillimprove.

I.I

NTRODUCTION

Tomographictransforms,suchastheRadontransform,can

beusedtoimplement“trackbeforedetect”processorsfor

thesimultaneousdetectionandtrackestimationofverysmall

entwork[1],weevaluatedtheexperimental

behaviorofthisapproachforthedetectionofsmalltargets

(periscopes)embeddedinrealhighresolutionseaclutter.

Insuchacontext,theRadontransformisanon-coherent

integrationstrategyalongthetarget’sexactpathintherange

ecifically,weproposedthe

useofawithinscan“trackbeforedetect”

antennarotationoperationmodewasusedtohavealong

dwellofaveryhighresolutionradarbeamoverthetarget.

Thecontinuouswavelinearfrequencymodulated(CWLFM)

radarused,transmittedabandwidthof1GHzandtherefore

hiscircumstances,

evenslowtargetsmoveacrossrangecellboundariesduring

thecoherentprocessinginterval,givingrisetolinearstructures

inarange-timerepresentation.

TheRadontransformisideallysuitedforthedetectionand

parameterestimationofthelinearstructuresintherange-

ctissupportedbythesuccessful

applicationofthistransforminthefieldsofradarsignalpro-

cessing,gh

transformbased“trackbeforedetect”systemdescribedin[2]

ortheshipwakedetectionsystemsproposedin[3],[4],[5],

aregoodexamplesthatclearlydemonstratethepotentialof

thisapproachforthedetectionoflinearstructuresinnoise.

Moreover,theoptimaldetectorofstraightpatternsinnoise

n

bedemonstratedwithinadetectiontheoryframework[6].

Radontransformbasedsystemsdatebacktothefirstyears

ofthedevelopmentof“trackbeforedetect”approaches[7].

However,severalfactsmakethemareasonablechoiceeven

whentakingintoconsiderationrecentcontributionstothefield,

suchasmaximumlikelihoodprobabilisticdataassociation

978-1-4244-2871-7/09/$25.00 ©2009 IEEE

(ML-PDA)[8],histogramprobabilisticmulti-hypothesistrack-

ing(Histogram-PMHT)[9],orparticlefiltering[10].Firstly,

despitethebruteforcenatureofRadontransformtechniques;

theirrelativecomputationalcostisgoodandaffordablefor

ly,theyareinherentlyrobust

y,noexplicitassumptions

aremadeonthestatisticalpropertiesofclutterortargets,

andthereforethereisnoperformancedegradationdueto

mismatchesbetweenmodelinghypothesisandoperativecon-

acommonproblemwhenmodelinghigh

resolutionseaclutter.

Inthisworkweproposeanovelmodificationofthebasic

directRadontransformtoperformacoherentintegrationof

theenergyoftargets,thatshouldimprovetheperformanceof

Radontransformbased“trackbeforedetect”systems.

II.W

ITHIN

S

CAN

R

ADON

T

RANSFORM

B

ASED

T

RACK

B

EFORE

D

ETECT

Highresolutionrange-timeimagesarecreatedbycolumn-

wisestorageofanumberofsuccessiverangeprofi

highresolutionradarsystemisworkinginspotlightmodeor

inaslowantennascanningoperation(withalongenough

coherentprocessinginterval)smalltargetswillhaveassociated

ialvelocity

ofeachtargetdeterminestheslantoftheline-likefeaturein

licationoftheRadontransform

toahighresolutionrange-timeimageisagridbased“track

beforedetect”techniqueinwhichallthepossibletargetpaths

lsoberegardedasa

convenientnon-coherentintegrationstrategyalongthetarget

exactpathintherange-timerepresentation,andtherefore

efficientlyusingthetarget’ditions

mustbemetforthiswithinscanapproachtobesuitable:

1)Thedwelltimeorcoherentprocessingintervalmustbe

longenough.

2)Therangeresolutionoftheradarmustbehigh.

Thisapproachwasevaluatedin[1]witha15cmrange

resolutionradarsystemandintegrationtimesof256ms.

TheRadontransformofacontinuoustwodimensional

functiong(x,y)canbedefinedas[11]: