Baseline correction using asymmetrically reweighted penalized least squares smoothing.

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Computational Science

Background:

• Penalized least squares methods are common for spectral baseline correction.
• Current methods may assign inappropriate weights to signals near the baseline, affecting accuracy.
• Noise distribution above and below the baseline necessitates balanced weighting.

Purpose of the Study:

• To develop an improved weighting scheme for penalized least squares baseline correction.
• To enhance the accuracy of spectral baseline correction and peak height estimation.
• To address limitations of existing methods in handling noise distribution.

Main Methods:

• Proposed a new weighting scheme utilizing the generalized logistic function.
• Implemented iterative noise level estimation and weight adjustment.
• Validated the method using simulated and measured Raman spectra.

Main Results:

• The proposed generalized logistic function-based method demonstrated superior performance.
• Outperformed existing methods in both baseline correction and peak height estimation.
• Effective in handling noise distribution above and below the estimated baseline.

Conclusions:

• The novel weighting scheme offers a significant advancement in spectral baseline correction.
• Provides more accurate peak height estimation compared to traditional approaches.
• Applicable to various spectral analysis tasks requiring robust baseline removal.