Finding Optimum Wavelet Decomposition Level Using Autocorrelation

Abstract

Wavelet transform is used to carry out time-frequency analysis including filtering of noises and reconstruction of the signal. Wavelet decomposition at a low level for denoising of signal is not very impactful in reducing noise. Alternatively, a high-level of wavelet decomposition may eliminate useful information carried by the signal. This study hinges on finding the optimum level of wavelet decomposition when performing denoising using wavelet transform. Optimum decomposition level is estimated based on parameters: mean and autocorrelation of residual in wavelet denoising. The importance of the paper is significantly the reduction of noise in the received signals whilst maintaining sufficient signal quality to allow the information carried by the signal to be read.