ABSTRACT
Human emotion recognition has evident and feasible benefits. It is possible to recognize human emotions with the help of the brain-computer interface. Data is recorded in multiple sessions for each topic at different dates and times. The recorded signal was first filtered with a band pass filter. The bandwidths selected for these signals are 8–12 Hz and 12–30 Hz, the alpha and beta bands of excitation and valence, respectively. This reduces the amount of noise that can occur in a relaxed position. Natural choices based on various basic emotions such as anger, joy, fear, curiosity, sadness, surprise, disgust, and acceptance are emotion recognition methods. However, this study maps emotional valence to arousal. This method has advantages in terms of suitability and simplicity over natural selection. The primary purpose of this study is to seek the modality of detectable stimuli and to quantify and manipulate them through an optimized number of electrodes that are effectively localized to the area responsible for audio-video processing. A test protocol is applied to the stimulus set, and signal changes in the brain are recorded. By filtering this signal, we removed the noise and analyzed the signal to calculate the signal's spectrum. Then the classifier is used. Modality, arousal, and valence classifications based on alpha and beta band performance were observed with a classification rate of 85.6%. The goal was to detect emotions rather than a modality, making it more challenging to classify modality. The modality classification rate is 82%, achieved by an optimized selection of principal components. The results showed that the arousal and valence categories were ranked higher than the modality. The results obtained are close to previous studies in this area but can be improved by choosing the appropriate electrode positions and classification methods.
