Publication

Abstract : To optimize the frequency recognition of SSVEP signals, various methods were proposed. The optimal frequency recognition methods follow the use of reference signal. Despite its efficiency, an issue identified is inaccurate detection of frequency since the reference signal used is the sine–cosine waves which is pre developed and this seems to lack the exact features present in the original electroencephalogram (EEG) data. To address this inaccuracy of frequency recognition, we propose to advance the reference signal by extricating the basic features (electrical activity of the brain) present in the EEG signals. This study also uses a spatial filtering mechanism to separate the task related activities called task related component analysis (TRCA). Having this TRCA has base, the proposed work functions on recording multiple sets of EEG data called, Multiset TRCA (M-TRCA). This streamlines the reference signal by removing the features that are common in the data sets. Also, an Ensemble M-TRCA that integrates the spatial filters to produce high speed and improved accuracy on frequency recognition is experimented. The proposed M-TRCA and Ensemble M-TRCA methods are compared to the existing TRCA and Ensemble TRCA methods under various parameters.