1M.Sc. Student, Department of Biomedical Engineering, Shahed University, Tehran, Iran.
2Associate Professor, Department of Biomedical Engineering, Shahed University, Tehran, Iran.
Multi-channels Electroencephaloram (EEG) needs a long preparation time for electrode installation. Furthermore, using a large number of EEG channels may contain redundant and noisy signals which may deteriorate the performance of the system. Therefore, channels reduction is a necessary step to save preparation time, enhance the user convenience and retain high performance for an EEG-based system. In this study, we present a simple and practical EEG-based emotion recognition system by optimizing the channels number based on two different Common Spatial Pattern (CSP) channel reduction methods. We applied feature extraction based on the Fast Fourier Transform (FFT) algorithm and classification method based on the Support Vector Machine (SVM) and K-nearest neighbor (KNN) which make our proposed system an efficient and easy-to-setup emotion recognition system. According to experimental results, the proposed system using small number of channels not only does not increase the error of the system, but also improves the performance of the system compared to the use of total number of channels.