Automated algorithm for baseline subtraction in spectra

A method for the automatic removal of a baseline in various types of spectra is proposed. The local values of the second derivative are used to distinguish between peaks and background, and the algorithm requires no user input. The performance is arguably better than all others tested in low-noise environments with varying baseline intensities, and comparable with the best routines in the case of spectra with a low signal-to-noise ratio. Copyright (C) 2010 John Wiley & Sons, Ltd.