A CWT-based SSVEP classification method for brain-computer interface system

Continuous wavelet transform (CWT) is used in this paper for steady-state visual evoked potential (SSVEP) detection in a brain-computer interface (BCI) system. The developed BCI system is designed for the remote control of humanoid robot through wireless sensor networks (WSN). A new CWT-based feature extraction method is presented and the whole framework of the BCI system is described. We investigated the feature extraction perfomance for different kinds of mother wavelets. Performance comparison was also conducted between CWT and fast Fourier transform (FFT). The experimental results show that the CWT-based method outperforms the FFT-based one in the SSVEP feature extraction scheme, specifically for short EEG segments. Moreover, the complex Morlet wavelet has significant superiority over several other mother wavelets in the CWT-based SSVEP feature extraction.