Wavelet Analysis Based Classification of Emotion from EEG Signal

Emotions are the most fundamental feature for non-verbal communication between human and machine. To extract the original expectation of mind, emotion recognition and classification is essential. But due to some complexities, the proper recognition of human emotion from Electroencephalogram (EEG) has become too much challenging. In this paper, we propose a system of emotion recognition from EEG signal based on Discrete Wavelet Transform. The most significant features (i) Wavelet Energy and (ii) Wavelet Entropy are calculated for detecting four different emotions namely happy, angry, sad and relaxed. Firstly we rearranged the prepossessed data properly by selecting the proper channel and sub-band. The extracted features are then trained in the K-Nearest Neighbor (KNN) algorithm to classify emotion separately. Our proposed method showed 78.7±2.6% sensitivity, 82.8±6.3 % specificity and 62.3±1.1% accuracy on the internationally authorized ‘DEAP’ database.