A　fine　spectral　negentropy　(ASNE)　method　based　on　empirical　wavelet　transform　(EWT)　is　proposed　to　solve　the　problems　that　it　is　difficult　to　determine　the　central　frequency　of　the　resonance　sideband　and　the　determination　of　the　bandwidth　is　susceptible　to　noise　when　extracting　fault　features　of　rolling　bearings.　The　proposed　method　constructs　a　filter　bank　by　using　the　characteristics　of　empirical　wavelet　filter　to　realize　the　scanning　filter　in　frequency　domain.　Then,　the　filtered　components　are　screened　by　combining　the　feature　of　spectral　negentropy　in　time　domain,　and　it　is　easier　to　detect　periodic　impulse　components　in　signals.　The　accurate　central　frequency　and　bandwidth　are　obtained　after　two　scanning　cycles.　Then　the　optimum　fault　feature　components　are　extracted　through　EWT,　and　the　fault　feature　information　of　the　bearing　is　finally　obtained　through　envelope　demodulation.　The　method　is　validated　by　the　experimental　signals　of　inner　and　outer　races　of　rolling　bearing.　The　results　show　that　the　method　quickly　and　accurately　determines　the　central　frequency　and　bandwidth　of　resonance　sideband,　and　effectively　extracts　the　fault　feature　information　of　inner　and　outer　races.　The　performance　is　better　than　that　of　the　Infogram　method.　The　proposed　method　overcomes　the　limitation　of　traditional　method　in　frequency　band　division　and　immunity　to　noise,　and　extracts　the　central　frequency　and　bandwidth　more　accurately.　©　2019,　Editorial　Office　of　Journal　of　Xi＇an　Jiaotong　University.　All　right　reserved.