多特征证据融合的遥感图像变化检测

《多特征证据融合的遥感图像变化检测》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、WANGMiIleta1．：Changedetectionusinghighspatialresolutionremotelysensedimagerybycombiningevidencetheoryandstructuralsimilarity559tionusinglinefeatures；Yuaneta1．，(2007)carriedoutbuildingwherechangedetectioncombiningimagespectrumandtextures；Wank=∑1-Img(4)eta1．(2008)proposedachan

2、gedetectionmethodusingtexturen4=妒l≤≤andspectralcorrelationcoefficient，ere．andkistheconflictdegreeofevidence．Eq．(3)iscalledtheInthisstudy,weproposeanovelremotesensingimageDempster’scombinationrule．changedetectionmethodcomprehensivelyconsideringedge，D—Sevidencetheoryisanefecti

3、vetoolforuncertaintyrea-textureandgradientchanges．Wecalculatethestructuralsimi．soning．Ithasmanysuccessfulapplicationsinthefieldsofre—1arityofthethree~aturesofbothtimephaseimages，createthemotesensing，e．g．，imageclassification，roadextractionandbasicprobabilityassignmentfunction

4、(BPAF)ofevidencethe—shadowdetectionfrommulticolorairbomeimages(Xiaoeta1．，oryforeachfeature，andimplementmulti-evidencefusion．2006；Dengeta1．，2007；Zhueta1．，2007)．Inthisstudy,itisWithdecisionrules，wegetthechangedimageareas．Compara—usedtocombineedge，textureandgradientchangesforti

5、veworkondiferentexperimentalareas，combinationsofmulti—featureimagechangedetection．changedevidenceandwithothermethodshasbeencarriedout．Itshowsthatourmethodpreventsefectivelythedetection2．2Structuralsimilarityerorsfromutilizingsinglefeatureandthusimprovesthedetec—Structuralsim

6、ilarity,whichwasfirstproposedbyWangettionprecision．口(2004)，hasalreadybeenusedinimagequalityevaluation．ThestructuralsimilarityoftwovectorsXandyisdefinedas：2METHODPRINCIPLEANDSTEPSSSIM(X，Y)[1(X，y)】·【c(，y)】_【s(X，y)】(4)Thismethodinvolvesimagevisualfeatureextraction，fea-whereture

7、histogramstatistics，structuralsimilaritycalculationofI(X，lr)：，c(，y)：2tXxt7y+C2featurevectors，andmulti—evidencefusion，etc．Wefirstintro．酸+善l专+C2oo；+C2，ducethebasicprincipleofevidencetheoryandstructuralsimi-larity,andthengivethemethoddetailedsteps．(，y)：，0+2·D-sevidencehe0ry，a11

8、d咖盯en1emea11，s啪darddevjaIj0D，D。Sevidencetheory1samathematicaltoolforuncerta