Threshold selection of wavelet denoising based on optimization algorithms

The wavelet threshold denoising is widely used to suppress the interference of noise and improve the precision of signal processing. The selection of appropriate thresholding values applied to the decomposition coefficients is very critical for the effect of noise filtering. The issue of threshold selection can be converted to optimization tasks by using different algorithms. In this study, the Aquila optimizer (AO), the gradient-based optimizer (GBO) and the modified grey wolf optimizer (GNHGWO) were utilized to optimize the threshold values. The well-known benchmark signals such as Blocks, Bumps, Doppler and Heavy sine were used for verifying the effect of different methods. The denoised signals were evaluated by two indices of signal-to-noise ratio (SNR) and root mean square error (RMSE). The simulation results on four benchmark signals have shown that the AO, G-NHGWO, GBO optimization algorithms used in this study have exhibited an encouraging effectiveness and practicability in threshold selection of wavelet denoising.