A FUNCTIONAL COMPARISON OF WINDOW SHRINKAGE SOFT THRESHOLDING TECHNIQUE WITH SOFT AND HARD THRESHOLDING TECHNIQUES FOR IMPROVED PERFORMANCE

Abstract

Thresholding techniques are pivotal in signal processing including denoising, and various estimation tasks. This article compares the functional performance of the window shrinkage soft thresholding technique with traditional soft and hard thresholding techniques. Among these techniques, soft and hard thresholding are widely used for denoising.Through theoretical analysis and empirical evaluation, the trade-offs in terms of computational complexity, denoising, and reconstruction accuracy was identified. Window shrinkage soft thresholding (WSST) function was developed based on inverted hanning window function. An Electrocardiographic signal, sampled at a frequency of 360Hz was captured from Institute of Technology-Beth Israel Hospital (MIT-BIH) online database for a duration of 5 seconds. The captured Electrocardiographic signal was loaded into a matlab environment and contaminated with a 50Hz powerline noise generated with matlab. Based on the new shrinkage function, decomposition level of 4 and daubechies 4 (db4) mother wavelet, the denoising of the contaminated Electrocardiographic signal was extensively performed using four different threshold estimation rules, namely sqtwolog, rigrsure, heursure and minimaxi threshold rules. The introduction of window shrinkage soft thresholding offers potential advantages in performance and the results demonstrate the advantages of its approach, particularly in adaptive noise environments, where it significantly outperforms conventional techniques indicating that the developed thresholding technique outperforms the existing ones as it possesses a power spectral density of -30.129 dB whereas the two existing ones possess power spectral density of -27.23 dB each which means that the developed technique effects better attenuation of the powerline noise.
A narrower window than the hanning window is recommended for future work as that will give better results comparatively.