I recently on the issue of research on wavelet denoising, see the theory of wavelet denoising is signal after wavelet transform energy concentrated so wavelet coefficient is larger, the noise energy spread and the wavelet coefficient is small, remove small amplitude of wavelet coefficients can be signal after denoising, but these assumptions for all seem to is a gaussian white noise, and I saw in another article in the English literature using adaptive wavelet threshold denoising can remove eye contour electricity of EEG signal of impulse noise, what is this principle? (but see again in other literature wavelet transform can well protect the pulse signal, then is it because they take the threshold of different reason or impulse noise and pulse signal have what different nature), in addition to correlation of wavelet transform or wavelet transform the signal can be related, and the noise after transformation trend of bleaching, so more than the time domain wavelet domain denoising, this to the correlation is not effective for all the noise? Small white problem is a little bit more, hope you can answer the great god, thank you.