A fully automated and model-free baseline-correction method for vibrational spectra is presented. It iteratively applies a small, but increasing, moving average window in conjunction with peak stripping to estimate spectral baselines. Peak stripping causes the area stripped from the spectrum to initially increase and then diminish as peak stripping proceeds to completion; a subsequent increase is generally indicative of the commencement of baseline stripping. Consequently, this local minimum is used as a stopping criterion. A backup is provided by a second stopping criterion based on the area under a third-order polynomial fitted to the first derivative of the current estimate of the baseline-free spectrum and also indicates whether baseline is being stripped. When the second stopping criterion is triggered instead of the first one, a proportionally scaled simulated Gaussian baseline is added to the current estimate of the baseline-free spectrum to act as an internal standard to facilitate subsequent processing and termination via the first stopping criterion. The method is conceptually simple, easy to implement, and fully automated. Good and consistent results were obtained on simulated and real Raman spectra, making it suitable for the fully automated baseline correction of large numbers of spectra.

The automated processing of data from high-throughput and real-time collection procedures is becoming a pressing problem. Currently the focus is shifting to automated smoothing techniques where, unlike background subtraction techniques, very few methods exist. We have developed a filter based on the widely used and conceptually simple moving average method or zero-order Savitzky-Golay filter and its iterative relative, the Kolmogorov-Zurbenko filter. A crucial difference, however, between these filters and our implementation is that our fully automated smoothing filter requires no parameter specification or parameter optimization. Results are comparable to, or better than, Savitzky-Golay filters with optimized parameters and superior to the automated iterative median filter. Our approach, because it is based on the highly familiar moving average concept, is intuitive, fast, and straightforward to implement and should therefore be of immediate and considerable practical use in a wide variety of spectroscopy applications.

We present here a fully automated spectral baseline-removal procedure. The method uses a large-window moving average to estimate the baseline; thus, it is a model-free approach with a peak-stripping method to remove spectral peaks. After processing, the baseline-corrected spectrum should yield a flat baseline and this endpoint can be verified with the χ(2)-statistic. The approach provides for multiple passes or iterations, based on a given χ(2)-statistic for convergence. If the baseline is acceptably flat given the χ(2)-statistic after the first pass at correction, the problem is solved. If not, the non-flat baseline (i.e., after the first effort or first pass at correction) should provide an indication of where the first pass caused too much or too little baseline to be subtracted. The second pass thus permits one to compensate for the errors incurred on the first pass. Thus, one can use a very large window so as to avoid affecting spectral peaks--even if the window is so large that the baseline is inaccurately removed--because baseline-correction errors can be assessed and compensated for on subsequent passes. We start with the largest possible window and gradually reduce it until acceptable baseline correction based on the χ(2) statistic is achieved. Results, obtained on both simulated and measured Raman data, are presented and discussed.

The two-point maximum entropy method (TPMEM) is a useful method for signal-to-noise ratio enhancement and deconvolution of spectra, but its efficacy is limited under conditions of high background offsets. This means that spectra with high average background levels, regions with high background in spectra with varying background levels, and regions of high signal-to-noise ratios are smoothed less effectively than spectra or spectral regions without these conditions. We report here on the cause of this TPMEM limitation and on appropriate baseline estimation and removal procedures that effectively minimize the effects on regularization. We also present a comparative analysis of TPMEM and Savitzky-Golay filtering to facilitate selection of the best technique under a given range of conditions.