Emotion Recognition from Physiological Signals Using Time-Frequency Analysis Methods

Emotions play an important role in the psychology and physiology of the human. It originates from the day to day experiences, people they interact with and the environment they deal with. In the current day of technological development, emotion recognition has got great attention, since it can contribute to the Human-machine interaction and for complete automation. Emotion recognition from physiological signals can help people with impairments in social interaction, communication and developmental disorders. This work tries to detect the human emotions from physiological signal, Electrocardiogram (ECG) using time-frequency analysis methods. It also provides a comparison of the performance of different time frequency methods of feature extraction. Discrete wavelet transform and Hilbert Haung transform are the time frequency methods used. For comparison two classification techniques Support Vector Machine and K-Nearest Neighbhour are also used. the emotions considered are happy and sad. Database was obtained from DECAF, a database of all physiological signals of different emotions.