An improved method of detecting and de-noising by the modulus maxima of wavelet transform

An improved de-noising method based on modulus maxima of wavelet transform is proposed. The theoretical calculation formula of noise variance at each scale and improved threshold method are proposed to realized the signal-noise separation. The results of simulation analysis of many non-stationary signal show that: the filtering accuracy can be significantly increased by the new method and the noise can be removed while the major details are remained well.