Modified energy based time-frequency features for classifying human emotions using EEG

Abstract

In this paper we summarize the emotion recognition from the electroencephalogram (EEG) signals. The combination of surface Laplacian filtering, time-frequency analysis (Wavelet Transform) and linear classifiers are used to detect the discrete emotions (happy, surprise, fear, disgust, and neutral) of human through EEG signals. EEG signals are collected from 20 subjects through 62 active electrodes, which are placed over the entire scalp based on International 10-10 system. All the signals are collected without much discomfort to the subjects, and can reflect the influence of emotion on the autonomic nervous system. An audio-visual (video clips) induction based protocol has been designed for evoking the discrete emotions. The raw EEG signals are preprocessed through Surface Laplacian filtering method and decomposed into five different EEG frequency bands using Wavelet Transform (WT). In our work, we used “db4” wavelet function for extracting the statistical features for classifying the emotions. A new statistical features based on frequency band energy and it’s modified from are discussed for achieving the maximum classification rate. The validation of statistical features is performed using 5 fold cross validation. In this work, KNN outperforms LDA by offering a maximum average classification rate of 78.4783 % on 62 channels and 73.6087% on 24 channels respectively. Finally we present the average classification accuracy and individual classification accuracy of two different classifiers for justifying the performance of our emotion recognition system.