Baseline correction for infrared spectra using relative absorbance-based independent component analysis

Infrared spectroscopy has important applications in fields such as materials analysis and chemical detection, while baseline correction is a key step in ensuring the accurate interpretation of spectral data. Uncorrected baselines can lead to deviations in absorption peaks, which affects the accuracy of both quantitative and qualitative analysis. When the absorption peaks of various components in a mixed gas overlap, the lack of reference baseline information in the continuous absorption band can result in significant errors during the baseline correction process. In this paper, we propose a relative absorbance-based independent component analysis (RA-ICA) algorithm to address this challenge. The algorithm first calculates the relative absorbance spectrum that excludes baseline information. Subsequently, based on the Beer-Lambert law and independent component analysis, it extracts independent components containing the absorption peak information of the components, allowing for fitting spectra that require baseline correction. Finally, a baseline model that combines polynomial curves and residuals is used to reconstruct the hidden baseline of the absorption band. Simulation and experimental results demonstrate that the baseline reconstructed using the RA-ICA method exhibits a significantly lower error compared to five other commonly used baseline correction methods and accurately preserves the detailed characteristics of the baseline.