Expanding the Capabilities of Portable Mapping in Macroscopic External Reflection FT-IR through a Targeted...

Reliable spectral quality is essential for extracting meaningful information from infrared reflectance data, particularly when using portable systems with limited scan numbers. This study presents a data-driven spectral enhancement workflow designed to improve the interpretability of portable macroscopic external reflection Fourier Transform Infrared (MA-rFT-IR) mapping systems developed by the Authors, operating in the near- and mid-infrared (NIR-MIR) ranges. Despite the growing use of reflectance imaging spectroscopy, limited attention has been devoted to the development of robust denoising strategies capable of minimizing noise and unwanted variability while preserving spectral quality and enabling more reliable and accurate data analysis. This study proposes a broadly applicable processing framework aimed at enhancing the efficiency and performance of reflectance-based spectral analysis. Denoising methods including Savitzky-Golay filtering and wavelet- and PCA-based denoising were tested and evaluated individually and in combination. Quantitative performance was assessed using arccosine similarity (ACOS) and derivative-based root-mean-square error (dRMSE) metrics across selected spectral regions of interest, with a derivative ACOS (dACOS) index applied to monitor band-shape variations. The evaluation results were integrated through Pareto analysis to identify the optimal trade-off between noise reduction and spectral-feature preservation. Application of the proposed approach to a multilayered painting mock-up demonstrated that the enhancing spectral data workflow preserves key diagnostic features revealing subtle spectral bands. Furthermore, applying multivariate curve resolution-alternating least-squares (MCR-ALS) to the denoised data enabled chemically meaningful separation of complex overlapping signals, improving the interpretability of compositional information compared with traditional denoising methods and data processing. The workflow strengthens the analytical reliability of low-scan reflection-mode data and provides a transferable framework for optimizing denoising strategies in portable infrared applications.