HREELS signal processing via wavelets

Abstract

AbstractIn spectroscopy, the recorded spectra can often be modelled as the noisy convolution product of an instrumental function with the ‘true’ signal to be estimated. Such models have often been used for high‐resolution electron energy‐loss spectroscopy (HREELS). In this article, a new method is suggested to estimate the ‘true’ HREELS signal, i.e. the original electronic diffusion function with ‘true’ peak intensities. Our method relies upon the use of wavelets that, because they exhibit simultaneous time and frequency localization, are well‐suited for signal analysis.Firstly, a wavelet shrinkage algorithm is used to filter the noise. This is achieved by decomposing the noisy signal into an appropriate wavelet basis and then thresholding the wavelet coefficients that contain noise. This algorithm has a particular threshold related to frequency and time.Secondly, the broadening due to the instrumental response is eliminated through a deconvolution process. This step mainly rests on the existing relation between the Lipschitz regularity of the signal and the decay with scale of its wavelet coefficients and on least squares.The efficiency of this technique is highlighted by comparing the results obtained with those provided by other published methods.This work is the second in a series of three papers in this issue. The first one presents background knowledge on the wavelets required to understand the estimation methods. The third paper explores the application of wavelet filtering and deconvolution techniques to x‐ray photoelectron spectroscopy. Copyright © 2004 John Wiley & Sons, Ltd.