简介:Satelliteremotesensingofinlandwaterbodyrequiresahighspatialresolutionandamultibandnarrowspectralresolution,whichmakesthefusionbetweenpanchromatic(PAN)andmulti-spectral(MS)imagesparticularlyimportant.TakingtheDaquekousectionoftheQiantangRiverasanobservationtarget,fourconventionalfusionmethodswidelyacceptedinsatelliteimageprocessing,includingpansharpening(PS),principalcomponentanalysis(PCA),Gram-Schmidt(GS),andwaveletfusion(WF),areutilizedtofuseMSandPANimagesofGF-1.TheresultsofsubjectiveandobjectiveevaluationmethodsapplicationindicatethatGSperformsthebest,followedbythePCA,theWFandthePSintheorderofdescending.Theexistenceofalargeareaofthewaterbodyisadominantfactorimpactingthefusionperformance.Meanwhile,theabilityofretainingspatialandspectralinformationsisanimportantfactoraffectingthefusionperformanceofdifferentfusionmethods.ThefundamentaldifferenceofreflectivityinformationacquisitionbetweenwaterandlandisthereasonforthefailureofconventionalfusionmethodsforlandobservationsuchasthePStobeusedinthepresenceofthelargewaterbody.Itissuggestedthattheadoptionoftheconventionalfusionmethodsintheobservingwaterbodyasthemaintargetshouldbetakenwithcaution.Theperformancesofthefusionmethodsneedre-assessmentwhenthelarge-scalewaterbodyispresentintheremotesensingimageorwhentheresearchaimsforthewaterbodyobservation.