Method
The proposed eyeblink detection algorithm first enhances
relevant features in the EGG and then determines
whether an EOG complex is an eyeblink or not via a composite
matching score. First a nonlinear erosion filter is
used to enhance troughs in the EGG signal that may be
possible eyeblinks. Then a differentiation filter is used to
reduce baseline drift. Finally, median and threshold filters
are utilized to suppress noise in the EGG signal.