多光譜遙感圖像配準與融合方法研究.pdf

1、南京理工大學碩士學位論文多光譜遙感圖像配準與融合方法研究姓名：司美玲申請學位級別：碩士專業(yè)：模式識別與智能系統(tǒng)指導教師：朱近；夏德深201103AbSt隨ct碩士論文AbstractMultispec砌imageh弱abulldantspectraJinfomlationwllichcreatesfIaVora【bleconditions南rreco印izingtlleeanhIsborder鋤dsurface，S0ith舔a謝de姍g

2、eofapplications鶴VefyhportamimageIntllepaper，meregisteringand血singmethodsofmulti—spec訓鼻e_motesensingimagearediscusscdAtfir％t11ispaperi砷的duCes咖嗆ecomerdetectionalgoritlullswtlichapplyt0meco礎mon莎ayimagesTheyareHarris，SUSANa1

3、1dCSScomcrdetectionoperal[0rT1lenmeimageregiS仃ationb囂edonSIFTisstudiedindetailiIlmispaperf0urapproachesareuSedinre西ste訂ngofmultispec刪remotesensingimageaboveTheexperimem面resultsdemo碰艦tet|lattlleeXl：ractedfeaturepointsoft1

4、1eH耐salgoritllmareScarCeaIldthee舭tedfea：tIlrepoimsofttleSUSANa190ritllmareinsensitiVet0coIItrastSo仕屺featl】repoin_tsofmatcllingcanhardlyfinishtheimageregis臼『ationACSSandSIFTalgorithmc觚ex廿actmany龜m眥pointsintheindiVidualban

5、d硫ageBotllmetllodsc鋤beVe巧goodtocompleteimagere西s幢a=tiont0multispec仃alremotesensiIlgimagewbjchh勰littledifjEerenceintllerayscaJearldspec乜albandsButthematchinge任Iectisnotgoodt0multi—spectralremotesensingimagewhichhasmuchdi僚

6、renceinmerayscaleaIldspectralbaIldsAccordillgtoMultispectralremotesensingimageWbjchischaracteriSticsoftheimageswithdiff宅rent罌ayleVelpropertiesaIldsubpixelpositionslliR，tllispaperproposesanewmetIlodbasedonlocal孕ayeXtrem呦m

7、etllodFirstlyitisneedtofindsmallarc勰contaillinggrayex仃em啪intIlereferellceimagea11dsaVetheare觴Detenllinet0alsocontainloc越伊ayex仃em岫intheregistr锨ionimageornotIfthereis，weke印theare釓0relSeitiseliIIliI埔艙dSccondlyⅡleex仃鋤a1re百on

8、s誦Ubes曲cefitting訛polynonlialLastly，extremepointsofthesmallareaarecalculatedrespectiVely鷸fea：curepoimsforregis協(xié)撕onIn“spaper，itisexperimem、析tllrcalandsimul紕edmulti—spec仃alimagesTheresultsshowmat幽eproposedapproachhaStheadV鋤

9、tagesofsimplealgorithmand900dprecisionOn恤b嬲isof婦ageregistration，tllispaper出s0iIl仃1)duceshage凡sionalgo訓lIllImagefhSionbeillganimportam印plicationofiIllageregis仃ation，tllispaperstudiesmultib鋤dimagemsionbaSedonwavelet仃ansfo肌

10、強dtllemsionofPaIlchromaticandMulti—spec砌ilnagcbaSedon也e口aditional、vavelet仃觚sfonn、HIS戧lllsf0蛐鋤dtll：ecombinationmemodofHISalldwavelet岫formTheexp麗mental托suhsdemons仃atetIl砒凡sionreSuhistllebestt0tllemuhibandiIIlagewhenlow骶que