A hybrid approach to emotion recognition system using multi-discriminant analysis & k-nearest neighbour

Emotion recognition system using electrocardiogram (ECG) has received considerable attention recently in the area of human computer interaction (HCI). Our work in this paper is an attempt towards developing an emotion recognition system that would classify emotions effectively into four emotional states: joy, anger, sadness and pleasure. The contributions of this paper is summarized in three fold: Firstly, we extract statistical features through digital filter. Secondly, to extract spectral features such as power and entropy features, we decompose and reconstruct the ECG signal through empirical mode decomposition (EMD) and apply Hilbert Huang transform (HHT) as well as discrete Fourier transform (DFT) to the intrinsic mode functions (IMFs). Finally, the effectiveness of our proposed hybrid scheme is demonstrated through experimental results in terms of various performance measures.