An Improved Wavelet Thresholding Denoising Method with Exponential-threshold Function Based on Particle Swarm Optimizer

摘要

Wavelet thresholding Denoising is one of main methods to eliminate noises. Via selecting proper threshold value and utilizing nonlinear methods with threshold functions to process wavelet coefficients, the optimum de-noising effect could be obtained in the sense of mean square deviation. Exponential-threshold function is widely used in wavelet thersholding denosing. However, how to select exponents is still an open problem. In this paper, the PSO technique is introduced to select the exponents of the prevalent exponential-threshold function with SNR as index function. Simulation results indicate that adopting the exponential-threshold function which is processed by optimization algorithms, better filtering effect could be acquired than the classical soft-, hard- and threshold methods.