imcorr软件使用方法.docx

上传人:b****0 文档编号:17212002 上传时间:2023-07-23 格式:DOCX 页数:20 大小:25.34KB
下载 相关 举报
imcorr软件使用方法.docx_第1页
第1页 / 共20页
imcorr软件使用方法.docx_第2页
第2页 / 共20页
imcorr软件使用方法.docx_第3页
第3页 / 共20页
imcorr软件使用方法.docx_第4页
第4页 / 共20页
imcorr软件使用方法.docx_第5页
第5页 / 共20页
imcorr软件使用方法.docx_第6页
第6页 / 共20页
imcorr软件使用方法.docx_第7页
第7页 / 共20页
imcorr软件使用方法.docx_第8页
第8页 / 共20页
imcorr软件使用方法.docx_第9页
第9页 / 共20页
imcorr软件使用方法.docx_第10页
第10页 / 共20页
imcorr软件使用方法.docx_第11页
第11页 / 共20页
imcorr软件使用方法.docx_第12页
第12页 / 共20页
imcorr软件使用方法.docx_第13页
第13页 / 共20页
imcorr软件使用方法.docx_第14页
第14页 / 共20页
imcorr软件使用方法.docx_第15页
第15页 / 共20页
imcorr软件使用方法.docx_第16页
第16页 / 共20页
imcorr软件使用方法.docx_第17页
第17页 / 共20页
imcorr软件使用方法.docx_第18页
第18页 / 共20页
imcorr软件使用方法.docx_第19页
第19页 / 共20页
imcorr软件使用方法.docx_第20页
第20页 / 共20页
亲,该文档总共20页,全部预览完了,如果喜欢就下载吧!
下载资源
资源描述

imcorr软件使用方法.docx

《imcorr软件使用方法.docx》由会员分享,可在线阅读,更多相关《imcorr软件使用方法.docx(20页珍藏版)》请在冰点文库上搜索。

imcorr软件使用方法.docx

imcorr软件使用方法

IMCORRRelease1.1

$Date:

2000-03-0214:

12:

28-07$

倒不是试验参数,读数据成了最大的问题

(Note:

Thisfile,themakefiles,andacoupleofotherdetailsaboutthe

distributionhavebeenslightlyeditedandputunderrevisioncontrolbyBruce

Raup.Thesoftwareisfunctionallyunchanged(exceptforoneminor

compile-timeoption),butthereleasenumberisnolonger1.0.Ifyouhave

commentsorsuggestionsforchangesoradditions,pleasesendthemtoB.

RaupandT.Scambos.)

NotesonusingIMCORRimage-to-imagegreylevelcorrelationsoftware

T.Scambos

(1)

M.Fahnestock

(2)

R.Bindschadler(3)

B.Raup(4)

(1)NationalSnowandIceDataCenterCIRESCampusBox449;154030thSt.

UniversityofColoradoBoulderCO80309-0449USA

vox:

303/492-1113fax:

303/492-2468net:

teds@icehouse.colorado.edu

(2)JointCenterforEarthSystemScience,DepartmentofMeterology,

UniversityofMaryland,CollegePark,MDUSA

voice:

301/405-5384net:

mark@firn.gsfc.nasa.gov

(3)Code971NASA/GoddardSpaceFlightCenterGreenbeltMD20771USA

Voice:

301/286-4717or301/286-2142Fax:

301/286-2717

Internet:

iceflow@laural.gsfc.nasa.gov(IP:

128.183.47.79)

(4)NationalSnowandIceDataCenter,CIRES;CampusBox449;154030thSt.

UniversityofColoradoBoulderCO80309-0449USA

Phone:

303-492-8814fax:

303-492-2468net:

braup@kryos.colorado.edu

 

INTRODUCTION

Thefollowingdocumentationisintendedtohelpinterestedusersacquire

andbeginusingIMCORR,asimpleimagecross-correlationprogramwhichcan

beusedtofindoffsetsoffeaturesinapairofimages.Thissoftware

usessubroutines【子程序,随源程序的原始码】foundwithinLAS5.0,anextensiveimageprocessing

packagewritteninFORTRANatNASA/Goddard.IMCORRconsistsofthese

subroutinesandasimpleCroutinethatactsasafrontend.Allthe

subroutinesrequiredfordisplacementanalysisarewithinthedirectory

associatedwiththisdocumentfile.

IMCORR

IMCORRtakestwoimagesandaseriesofinputparametersandattemptsto

matchsmallsubscenes(called'chips'【小窗口】)fromthetwoimages.Theprogram

usesafastfouriertransform【FFT快速傅里叶变换】-basedversionofanormalized

cross-covariance【正交互协方差】method(seeBerenstein,1983;alsoScambosetal.,1992).

Themostcommonuseofthistypeofalgorithminimageprocessingisto

accuratelylocatetie-point【连接点,栓点】pairsintwoimagestocoregisterthem.

However,iftheimagesarealreadycoregisteredbyothermeans,the

algorithmmaybeusedtofindthedisplacementsofmovingfeatures,

providedthatthefeaturesshowlittlechangeintheirappearance,andthat

themotionisstrictlytranslational【平动,直线】.IMCORRtakesasinputtheimage

namesandsizes,parametersdeterminingsearchchipsize,referencechip

size,gridspacing,andoutputfilename.【全部参数】Further,presetoffsetsofsearch

chipcentersmaybespecified,andsubareasofthefullimagefilesmaybe

usedtorestricttheareaoverwhichIMCORRattemptstofinddisplacements.

AteachofthegridpointsIMCORRcalculatesacorrelationindexforevery

locationatwhichthereferencechipwillentirelyfitwithinthesearch

chip.IMCORRtakesthecorrelationvaluesinthevicinityofthebest

integer-pixel【整数像素】matchandinterpolatesapeakcorrelationlocationto

sub-pixelprecision.Theprogramreturnsafilecontainingthelocations

ofthegridcentersforthereferencechips,thedisplacementsrequiredto

bestmatchthechippairs(orindicatesthatnonecouldbefound),and

severalqualitycontrolparametersthatmaybeusedtoevaluatethe

validityofthematch.Weusethisprogramtomeasureglaicervelocities;

however,thesameprogrammaybeusefulforotherapplications.

Itisimportantforthereadertounderstandthatallofthecorrelation,【相互关联】

peakfinding,anderrorestimationisdonebytheFORTRANsubroutinesfrom

LAS,whichwaswrittenbyothersatNASAlongbeforetheauthorsofthis

documentwereinvolvedinimageanalysis.WehavesimplyaddedtheCcode

whichmakestheuseoftheseroutinesmorestraightforwardandautomated

forvelocity-mappingapplications.

Apaperdescribingthedetailsofthistechniqueandthepreprocessingof

imagesrequiredtooptimizethedisplacementmeasurementsispublishedin

RemoteSensingofEnvironment(Scambosetal.,1992;seebibliography).We

willgladlyprovideapreprintuponrequest.

GETTINGTHESOFTWARE

Thelatestdistributionforimcorrisat

http:

//www-nsidc.colorado.edu/NSIDC/ANTARCT_VELOC/imcorr.html.

LOADINGTHESOFTWARE

Transfertheentirecontentsofthe'distribution'directorytoyour

workstation(thetutorialdiscussionbelowwillassumeyouhavetransferred

thefourtestimageswehaveincludedinthedirectory).Thisdirectoryis

roughly1.25Mbytes,andcontainsthefollowingfiles:

conv.y87testimagefile

conv.y89testimagefile

correlate.hHeaderfileforimcorr.c

cross.fFORTRANsource

esterr.fFORTRANsource

eval.fFORTRANsource

fang.y87testimagefile

fang.y89testimagefile%看样子有两组完整的实验数据

fft2d.fFORTRANsource

fitreg.fFORTRANsource

gcorr.fFORTRANsource

gnorm.fFORTRANsource

imcorr.cCsource

kvert.fFORTRANsource

makefile.LinuxmakefileforLinuxmachines%要用到的makefile

makefile.SGImakefileforSiliconGraphicsmachines

makefile.SUNmakefileforSunMicrosystemsmachines

README.txtthisfile

sums.fFORTRANsource

test1.outExampleoutputfile

test2.outExampleoutputfile

test3.outExampleoutputfile

TO_DO.txtlistofto-do's

COMPILE-TIMEOPTION

Beforecompiling,youcanuncomment(deletethe'#'character)thelinein

themakefilethatsays

#DEFS=-DCOMMANDLINE

Thiswillcauseimcorrtoprintthecommand-linethatwasissuedatthe

shellpromptasthefirstlineintheoutputfile,precededbya'#'

character.

【正式编译的时候,可以反注释(本来是注释的,删掉注释符号),但是如果不删掉也没关系吧?

只是在文件开头有所区别】

HOWTOCOMPILE

TheIMCORRsourcecodeisroughly65kilobytes.ItrequiresbothCandFORTRAN

compilers.Itcontainsnographicscommandsineitherlanguage,andshouldbe

relativelyeasytoporttoavarietyofsystems.WeuseSiliconGraphics

workstations;thesoftwarehasalsobeencompiledonSUNSparcstationsand

Linuxmachines.WehaveincludedmakefilefilesforSGI,Linux,andSUN

workstations.Togenerateanexecutablefilefromthesourcecode,type:

prompt>make-fmakefile.[SUNorSGIorLinux]imcorr,产生可执行文件

Onanysystem,toremovetheobjectfiles,type:

prompt>make-fmakefile.[SUNorSGIorLinux]clean

Ifyouwantorneedtomakeanychangestothemakefile,werecommend:

prompt>cpmakefile.[SUNorSGIorLinux]Makefile

prompt>viMakefile(makechangeswithwhatevereditoryoulike)

prompt>make

Bydefault,makeuses"Makefile"asinput,ifitexists.

Allofthesemakefilesproduceanexecutablefilenamedimcorr.

INPUTPARAMETERS

ThisIMCORRversiontakeseither5,10,or14inputparameters,【重载函数可默认设置】whichare

typedonthesamelineasthe'imcorr'commandseparatedbyspaces.Simply

typing'imcorr'andreturngivesalistoftherequiredinputparameters

foreachofthesethreeformats.

ListofInputParameters:

referenceimagefilename(generallytheolderimage)

searchimagefilename(generallythenewerimage)

numberofpixelsperlineinimages(nodefault)

numberoflinesinimages(nodefault)

outputfilename(doesnotneedtoexistpriortoarun)

searchchipsizeinpixels

(default:

64;mustbedivisibleby16,lessthanorequalto256;

entering'32'meansasearchchipof32x32pixels)

referencechipsizeinpixels

(default:

32;mustbedivisibleby16,lessthanorequalto128;

entering'32'meansarefchipof32x32pixels)

gridspacinginpixels

(default:

25;nolimitsexceptthoseimposedbytheimages;

gridsarealwayssquare;entering'25'meansthatthesearchand

referenceimagechipswillbecenteredonpoints25pixelsapart

inbothlineandsampledirections)

xoffsetofreferencechiprelativetosearchchipcenter(default:

0)

yoffsetofreferencechiprelativetosearchchipcenter(default:

0)

xoffsetofsubimagetobeanalysed(default:

0)

yoffsetofsubimagetobeanalysed(default:

0)

xextentofsubimage(default:

fullxextent)

yextentofsubimage(default:

fullyextent)

Thereferenceimagewillhavethesmaller,'referencechip',subimagesderived

fromit,atregulargridspacings,【网格间距】andthesechipswillbecomparedtolarger

'searchchip'subimagesderivedfromthesearchimage.Theimagesmustbethe

samesize,andasingle(sample,line)coordinatesystemwillbeusedforboth

ofthem.Inthefollowingdiscussion,itisassumedthattheimagesare

coregistered,i.e.,thatafeaturethatdoesnotmovewouldbefoundatthesame

(sample,line)coordinatesinbothimages.Regardingmaximumimagesize,

machinememoryisdynamicallyallocatedduringanIMCORRrun.Therefore,images

maybeverylarge;however,largerimageswillresultinslowerprocessingdue

tomemoryswapping.【内存交换】

Thexandyoffsetsofthesubimageanswerthequestion,"wheredoesthe

referencechipcomefrom?

".Forexample,iftheicehasmovedinthepositive

xdirection(totheright)by20pixels,thenthex-offsetshouldbe-20.

【offset说明了方向性】

Thecoordinatesystemhasxpositivetotheright,andypositivedownward:

-------------->x

|

|

|

|

|

|

V

y

Largesearchandreferencechipsizesslowtheruntimeconsiderably,anda

largereferencechipmaycauseproblemsifanydistortionofthefeatures

occursbetweenthetwoimages.Notethatifthesearchandreferencechipsare

derivedfromthesame(sample,line)gridcentersintheirrespectiveimages

(thedefaultcase),themaximumdisplacementofafeaturethatcanbemeasured

is:

(searchchipsize/2-referencechipsize/2)*C2,andthismaximumapplies

onlytoexactlydiagonalmotion.

Gridspacingparameterdeterminesthedensity

展开阅读全文
相关资源
猜你喜欢
相关搜索
资源标签

当前位置:首页 > 解决方案 > 学习计划

copyright@ 2008-2023 冰点文库 网站版权所有

经营许可证编号:鄂ICP备19020893号-2