Adaptive recognition and filtering of noise using wavelets

We combine wavelet transform and adaptive recognition techniques to introduce a filtering process able to analyze, categorize, and remove additive noise from experimental time series, without previous information either on the correlation properties of noise or on the dimension of the deterministic signal. The method is applied to a high dimensional delayed chaotic time series affected by additive white and colored noises. The obtained results show that the reconstruction of the signal both in real and in Fourier space is effective through the discrimination of noise from the deterministic part.