Wavelet denoising method for laser-induced breakdown spectroscopy

Abstract

The wavelet threshold denoising method is an effective noise suppression approach for noisy laser-induced breakdown spectroscopy spectrum. The wavelet threshold denoising method is influenced by several key issues such as the choice of wavelet, the choice of decomposition level, threshold selection, and the choice of thresholding functions. In this paper, the double threshold optimization models of semi-soft thresholding function are established firstly. Next, on the basis of grey relational analysis and Euclid closeness of fuzzy theory, a method of amending the double thresholds of semi-soft thresholding function is put forward. The performance of the proposed method is verified by analysis of both synthetic and observed signals. The limit of detection values are reduced by more than 50% and the signal to noise ratios are improved by a factor of two by using the proposed method.