DefinePK

Abstract

Facial electromyography is a modality for designing emotion recognition system which is gaining popularity as a human machine interface to control the devices. In this research, we analyse the Facial Electromyography (FEMG) signals for the six emotions namely anger, disgust, fear, happy, neutral and sad using the multirate features. Multirate signal processing is a technique which alters the rate of the discrete-time signals, either by adding or deleting a portion of the signal samples. The advantages of such multirate features are that it increases the processing efficiency and reduces DSP hardware requirements. Twenty subjects took part in this experimental study. Three multirate features are used to derive the significant features. Six emotions were identified by applying the multirate features as input to neural network models. Two network models namely Cascade Network and Fitting Network were used and compared to identify an efficient network for emotion identification. The performance of the networks identified the six emotions in the range of 78.56% to 98.72%.