Emotion Recognition based on EEG using IMF Energy Moment

Abstract

In this paper, we proposed a novel model for raising the classification accuracy rate of EEG emotion recognition by combining empirical mode decomposition (EMD) and wavelet transform to extract the energy moment of EEG. Based on wavelet transform α` θ `β and γ rhythms in this paper were extracted at left and right prefrontal lobe (AF3, AF4), frontal lobe (F3,F4) and parietal lobe (FC5,FC6). Intrinsic mode function (IMF) was analyzed and exacted based on EMD. furthermore, energy moment of IMF was analyzed and obtained. The support vector machine was used to assess the state of emotion which supports the music therapy. The result proved that it is feasible for EEG emotion recognition by using IMF energy moment.