Abstract:
An approach based on the using of the continuous wavelet transform (CWT) in derivative spectrometry (DS) is considered. Within the framework of the approach we develop a numerical differentiation algorithm with continuous wavelets for improving resolution of composite spectra. The wavelet-based derivative spectrometry (WDS) method results in best contrast in differential curves compared to the conventional derivative spectrometry method. A main advantage is that, as opposed to DS, WDS gives stable estimations of derivative in the wavelet domain without using the regularization. A wavelet shape and the information redundancy are of the greatest importance when the continuous wavelet transform is used. As an appropriate wavelet we offer to utilize the nth derivative of a component with a priori known shape. The energy distribution into scales allows one to determine a unique wavelet projection and in that way to avoid the information redundancy. A comparative study of WDS and DS with the statistical regularization method (SRM) is made; in particular, limits of applicability of these are given. Examples of the application of both DS and WDS for improving resolution of synthetic composite bands and real-world composite ones coming from molecular spectroscopy are given. © 2004 Elsevier B.V. All rights reserved.
