文摘
In this paper we present a new wavelet denoising (WD) enhanced principal component analysis (PCA) method (wPCA) to reduce the number of trials required for the efficient extraction of brain event related potentials (ERPs). First, the ERPs are extracted with wavelet transform, giving us an enhanced version of the raw data. Next, the principal components (PCs) with most of the total variance are considered to be part of the ERP subspace. Lastly, the ERPs are reconstructed from the selected PCs. Simulation and experimental results show that the wPCA method provides better performance than either WD or PCA method.