A wavelet denoising method based on improved threshold and autocorrelation
Ying Qian · 2018
In this paper, the noise of high frequency coefficients in wavelet decomposition is analyzed by autocorrelation function, which may determine which layer's wavelet high frequency coefficients to take part in wavelet reconstruction. The appropriate denoising threshold is selected according to noise in the signal. In considering of the advantages and disadvantages of hard threshold and soft threshold, an improved threshold function is designed. The wavelet high frequency coefficients of each layer are denoised By the improved threshold function and then reconstructed. Combining wavelet threshold denoising with wavelet decomposition and reconstruction, the denoising effectiveness is verified by MATLAB simulations, and the improved threshold function is proved to have better denoising effect than that of hard threshold and soft threshold.