Abstract: In traditional denoising methods associated with wavelet coefficients, wavelet coefficients on any scale are only slightly offset after wavelet transformation to a noisy signal. A wavelet coefficients correlation denoising method based on a cross-correlation function was proposed. A cross-correlation algorithm was used to calculate the deviation of various scales and the original noisy signal, and to analyze the deviated scale, thus obtaining an accurate mutant signal. In order to remove the small number of noisy signals in the vicinity of the mutation signals, a critical threshold denoising method was used to do threshold processing with the scale signal after correlation analysis, following which an accurate useful signal was reconstructed.