简介:Empiricalmodedecomposition(EMD)isanewsignaldecompositionmethod,whichcoulddecomposethenon-stationarysignalintoseveralsingle-componentintrinsicmodefunctions(IMFs)andeachIMFhassomephysicalmeanings.ThispaperstudiesthesingletrialextractionofvisualevokedpotentialbycombiningEMDandwaveletthresholdfilter.ExperimentalresultsshowedthattheEMDbasedmethodcanseparatethenoiseoutoftheeventrelatedpotentials(ERPs)andeffectivelyextracttheweakERPsinstrongbackgroundnoise,whichmanifestedasthewaveformcharacteristicsandrootmeansquareerror(RMSE).