Step-down approach for wavelet thresholding

Abstract

Wavelet thresholding is one of the effective denoising methods for signal processing. It eliminates the noises by removing wavelet coefficients less than the specified threshold. The existing methods set their threshold values based on the variance or length of the signal. However, these approaches are not able to consider the overall trend or the characteristics of data effectively. In this paper, we proposed the step-down denoising for wavelet thresholding. The step-down approach is a data reduction method that defines the threshold value by calculating the order statistics of wavelet coefficients. The suggested method is applied to four types of sample data with various levels of signal-to-noise ratio (SNR). To evaluate the performance of this approach, the comparison of the denoising results in terms of plotting and average root mean square error (AMSE) is carried out.