FREQUENCY RECOGNITION IN SSVEP- BASED BCI USING EMD CCA TECHNIQUE
Abstract
Steady-State Visual evoked Potential is periodic evoked signals buried into non-stationary EEG recorded Signal when subject focuses on visual target. Accurate detection of SSVEP Signals is challenging task in EEG based Brain-computer Interface System. The presence of artifacts and spontaneous brain activity during EEG Signal recording may deteriorate the performance of the SSVEP Based BCIs System. Therefore, the alternate option is to extract the few frequency sub-bands in which SSVEP Signal Prominent instead of using the entire band of frequency to minimize the unrelated brain activity and artifacts. This paper investigate how Empirical Mode Decomposition (EMD) method decompose the recorded EEG Signal into oscillating component called Intrinsic mode function(IMFs).The IMFs accountings for SSVEP components are selected for Target frequency identification. Finally the detection accuracy is measured by applying canonical correlation Analysis (CCA) on selected IMFs component. The result shows that the detection of SSVEP Signal using the EMD-CCA technique gives a better result as compared CCA technique applied to the entire raw EEG Signal.