Number recognition using salient features of beta rhythmic EEG signal

Development of visual recognition system using electroencephalogram (EEG) signal is the most challenging task in the present computational neuroscience. The goal of this work is to recognize different numbers using salient features of EEG signal. Using data acquisition system, EEG signal is recorded applying visual stimulation. Visual stimulation of different numbers produces different patterns in human brain that are basically reflected in EEG signal as well as its different rhythms. EEG signal with different rhythms is collected using Microsoft PowerPoint and significant features are extracted using frequency domain and time-frequency domain analysis. It is observed that the analysis of beta rhythmic EEG signal gives better results than other rhythms for number recognition. Wavelet transform analysis of beta rhythm can't distinguish different numbers whereas fast Fourier transforms and power spectral density can provide unique pattern for each number. Finally, power spectral density provides superior features than fast Fourier transform for number recognition. Beta rhythmic EEG signal based number recognition has direct impacts in neurologically disorder persons as well as in robotics.