A spatial constraint to model and extract texture components in Multivariate Curve Resolution of near-infrared hyperspectral images

Raffaele Vitale; Siewert Hugelier; Dario Cevoli; Cyril Ruckebusch

doi:10.1016/j.aca.2019.10.028

A spatial constraint to model and extract texture components in Multivariate Curve Resolution of near-infrared hyperspectral images

Anal Chim Acta. 2020 Jan 25:1095:30-37. doi: 10.1016/j.aca.2019.10.028. Epub 2019 Oct 17.

Authors

Raffaele Vitale¹, Siewert Hugelier², Dario Cevoli², Cyril Ruckebusch²

Affiliations

¹ Laboratoire de Spectrochimie Infrarouge et Raman - LASIR CNRS - UMR 8516, Université de Lille, Bâtiment C5, F-59000, Lille, France; Molecular Imaging and Photonics Unit, Department of Chemistry, Katholieke Universiteit Leuven, Celestijnenlaan 200F, B-3001, Leuven, Belgium. Electronic address: rvitale86@gmail.com.
² Laboratoire de Spectrochimie Infrarouge et Raman - LASIR CNRS - UMR 8516, Université de Lille, Bâtiment C5, F-59000, Lille, France.

PMID: 31864628
DOI: 10.1016/j.aca.2019.10.028

Abstract

This article highlights the importance of properly taking into account spatial structures and features to better resolve near-infrared (NIR) hyperspectral images by Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS), especially when highly mixed components (in terms of spatial and spectral overlap) underlying the systems under study are dealt with. As in the NIR domain these components can explain both chemical properties and physical phenomena, their improved unravelling can therefore represent an alternative or a complement to more standard approaches for, e.g., spectral data preprocessing. These points will be illustrated through the comprehensive analysis of a complex real-world forensic case-study where texture characterization is crucial for the sake of a more appropriate resolution.

Keywords: Forensics; Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS); Near-infrared (NIR) hyperspectral images; Spatial constraints; Texture extraction.