The　ocular　movements　are　inevitable　in　electroencephalography　(EEG)　collection,　and　the　induced　ocular　artifacts　(OA)　have　larger　amplitude　than　the　other　noises.　The　OA　removal　effect　is　very　important　to　the　following　process.　Based　on　canonical　correlation　analysis　(CCA)　and　discrete　wavelet　transform　(DWT),　a　novel　method　was　proposed　to　recognize　and　eliminate　OA　in　EEG,　and　it　was　denoted　as　DWT-CCA.　Firstly,　DWT　was　applied　to　the　raw　EEG　and　electrooculogram　to　acquire　the　multiscale　wavelet　coefficients,　and　CCA　was　used　to　eliminate　their　auto　correlation.　Then　the　correlation　coefficient　was　selected　to　recognize　the　ocular　components,　and　the　corresponding　coefficients　were　set　to　zeros.　At　last,　the　inverse　algorithms　of　CCA　and　DWT　were　completed　in　sequence　to　obtain　the　clean　EEG.　Experiment　results　show　that　DWT-CCA　has　perfect　effect　in　some　aspects　such　as　root　mean　square　error　(RMSE)　and　signal　noise　rate　(SNR)　etc.