Preprocessing EEG signals for direct human-system interface

EEG-based direct human-system interfaces attempt to provide a direct brain activated control for quadriplegics and workers in high-workload environments. But significant practical obstacles due to individual differences, subject generated artifacts, noise and interference remain. Eyeblinks and eyeball movements artifacts cause severe interference in EEG recorded by forehead electrodes. This work concentrates on real-time removal of eye-blink and eyeball movement artifacts from the forehead EEG recordings. The proposed approach uses a fixed bandpass FIR filter to remove slow baseline drift. This is followed by a subject specific eyeblink threshold. The results obtained from a real-time implementation on a general-purpose computer are also presented