Investigation of a genetic algorithm based cubic spline smoothing for baseline correction of Raman spectra

Abstract

Raman spectral analysis has been seriously influenced by the undesired signals such as background from samples themselves or other interfering materials. Many mathematic algorithms have been proposed to eliminate the background of Raman spectra. However, these methods require appropriate parameters for the Raman spectral baseline correction. Therefore, we propose a genetic algorithm based cubic spline smoothing (GaCspline) method for baseline correction of Raman spectra in this paper. Genetic algorithm has been applied to choose the spectral wavenumbers which belong to a background in non-Raman characteristic peak channels. Then, these suspected background wavenumbers are fitted with cubic spline smoothing method. The simulated results demonstrate that the proposed GaCspline baseline correction method is better than asymmetric least squares and adaptive iteratively reweighted penalized least squares methods in background elimination. And, when the real Raman spectra are treated by the GaCspline method, the results indicate that this method can well handle the complex background and keep the Raman characteristic features as well.