Dual-Tree Complex Wavelet Transform-Based Feature Extraction For Brain Computer Interface

The dual-tree complex wavelet transform (DTCWT) is good at time-frequency analysis and has shift-invariance property. In this paper, we propose a feature extraction method based on DTCWT, which employs the DTCWT to reconstruct the brain computer interface (BCI) signals in each level and overcome the frequency aliasing in wavelet transform. The experimental dataset come from the BCI competition, the mutual information and classify accuracy are used as evaluation criteria. The results show that the DTCWT-based feature extraction method improves the mutual information and accuracy compared to others.