Publication

Abstract : Emotion analysis is an emerging field among current researchers. Emotion plays an important role in forming behavioural patterns in the human brain. A study was conducted in objective time space evolution of emotions. For emotion recognition, an EEG (electroencephalogram) data base was created from equal number of male and female subjects and named Amrita-emote database (ADB). The subjects were shown videos of various emotions and simultaneously the output EEG signals were recorded into ADB, which contained 500 samples of data. Our study had focused into 5 basic emotions, Viz. neutral, happy, disgust, sad and fear. The ADB was split into two different frequency bands of 12-35 Hz as high frequency band (HFB) and 1-8Hz as low frequency band (LFB) for time localized responses. The time and space evolutions were studied by segmenting and K-mean clustering. For each emotion, the frequency distribution contributes from four regions in the brain. The statistical analysis was done to find the average contribution for each emotion.