点击次数：

论文类型： 期刊论文

发表时间： 2012-01-01

发表刊物： ICIC Express Letters

收录刊物： EI、Scopus

卷号： 6

期号： 9

页面范围： 2377-2382

ISSN号： 1881803X

摘要： To estimate a signal from colored noises is a challenging problem because of the frequency overlap as well as the low signal-to-noise (SNR). In this paper, a novel adaptive wavelet method based on pre-whitening of colored EEG noises for single-trial evoked potential (EP) estimation is proposed. This method does not require prior information about the EPs. In particular, an adaptive weighted thresholding method is formulated using the least-mean-square error (LMSE) scheme to avoid manual selection of wavelet coefficients. For validation purposes, the proposed algorithm is assessed in its ability to extract brain EP corrupted by the colored EEG noises in which SNR values can be as low as -10 dB in real clinical environments. The experimental results show that this algorithm can successfully estimate dynamic changes in EPs, even under poor signal-to-noise ratio (SNR) conditions. ? 2012 ICIC International.