Challenges and spectra interpretability in textile sorting: NIR hyperspectral images and chemometrics

The accurate sorting and recycling of textiles are crucial for reducing waste and promoting sustainability, but the inherent complexity of textile materials presents significant challenges. This study explores the potential of Near-Infrared (NIR) hyperspectral imaging (HSI) in textile sorting, focusing on its capacity to address two key challenges: differentiating textiles with diverse chemical compositions and detecting minor components such as elastane in cotton fibers. We investigate the spectral characterization of natural, synthetic, and blended fibers, demonstrating how NIR-HSI, combined with complementary spectroscopic approaches, can provide both chemical composition and spatial distribution insights. The integration of portable spectroscopic techniques improved spectral data interpretation and models reliability. The study also highlights the challenges of detecting low concentrations of elastane, which complicate recycling processes. Through the application of multivariate chemometric techniques, such as Principal Component Analysis (PCA) and Multivariate Curve Resolution (MCR), we analyze the spectral data and identify patterns that may facilitate more effective textile classification. The results indicate that while NIR-HSI shows promise in textile sorting and recycling, the variability in spectral features due to textile treatments, environmental factors, and compositional differences remains a significant challenge. The study highlights the importance of the development of standardized spectral databases to combine with interpretation. Ultimately, the findings offer valuable insights into the feasibility of near-infrared (NIR) hyperspectral imaging as a tool for sustainable textile sorting and recycling.