Employing minimum distance classifier for emotion recognition analysis using EEG signals

摘要

Emotion recognition is a process of identifying human emotions using facial expressions, speech, physiological signals, etc. EEG signal has a low noise ratio, as they originate from brain and are not prone to social masking. We have selected the physiological EEG signal to recognise human emotions. The acquired EEG signal is processed, and various prominent features have been extracted on the basis of which the emotions are classified using the minimum distance classifier into four quadrants of valence and arousal. From the analysis, we have discussed how these different feature combinations and the different distances which are applied for emotion classification have been selected. From the analysis Manhattan distance is proved as the best for feature classification as it gives the accuracy of 97.5%.