以凸面单体边界为搜索空间的端元快速提取算法

《以凸面单体边界为搜索空间的端元快速提取算法》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、JournalofRemoteSensing遥感学报2010，14(3)—5口口石嗣j确—50o县o口Iq首3rH∞3口，5443SBAandN．FINDR．Fig．8illustratestherelationshipbetweenf2)BecaOuse9alle8ndm7em6bers5are4inth3eun2ionofal1bounda．55555555505O5thenumberoftrainingsamplesandtheratioofthetimecon—riesoftwo．dimensional‘‘cloudsofpoints”．thisunionofsumptionofN

2、—FINDRtoSBA．boundariescanbeusedassearchspaceinSBAalgorithm，and【ble4andFig．7showthatthetimecostsofSBAandallendmemberscanbefoundfromthisunion．N—FINDRa1lincreaserapidlywhilethenumberoftrainingr3)Withanincreaseofthenumberofendmembersortrainingsamplesincreases．andSBAisalwaysfasterthanN-FINDR．samples．SB

3、AhasgreateradvantagesinspeedthanN．FINDR．1'able4andFig．8showthattheratioofthetimeconsumptionREFERENCESofN．F1_NDRt0SBAisalwaysgreaterthanoneforanysizeofthesubimages．anditalwaysincreases．However,thisincreaseBajorskiP‘SchoRJandIentilucc1E．2004．Comparisonofbasis—vectorbecomesslowerandslowerwhenthenumbe

4、roftrainingsam—selectionmethodsfortargetandbackgroundsubspacesasappliedtosubpixeltargetdetection．SPIEProceeding，Algorithmsandpiesexceedsavalue，forexample5x10inthisexperiment．TechnologiesforMultispectral，HyperspectralandUltraspectralThisstudymeansthatSBAhasmoreadvantagesthanImageryXN．FINDRwhenmoret

5、rainingsamplesareusedforendmemberBoardmanJK／useFAandGreenRO．1995．Mappingtargetsigna—extraction．taresviapartialtmmixingofAVIRISdata．Summariesof5thAnnualJPLAirborneEarthScienceWorkshop，JetPropulsionLaboratory,Table4TimeconsumptionofSBAandN-FINDRfordiferentPasadena，CA，JPLPublicationnumberoftrainingsa

6、mplesBoardmanJ1993．AutomatingspectralunmixingofAVIRISdatausingconvexgeometryconcepts．Summariesof4thAnnualJPLAirbomeGeoscienceWorkshop，JetPropulsionLaboratory,Pasa-dena，CA，JPLPublicationBowlesJ，DanielM，GrossmanJ，AntoniadesJ，BaumbackMandPal—madessoP1998．Comparisonofoutputfromoasisandpixelpurltycalcu

7、lations．SPIEProceeding，ImagingSpectrometryIV,3438：148—156ChangCI．WuCC．LiuWMandOuyangYC．2006．Anewgrowingmethodforsimplex—basedendmemberextractionalgorithm．IEEETransactionsonGeoscienceandRemoteSensing．44(10)：28O4—2