Factorization of non-negative matrix for spectral mixing of time series and voxel-based machine learning: application to biodegradable films.

• Grant number: 24/03826-1
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Start date: July 01, 2024
• End date: June 30, 2025
• Field of knowledge: Agronomical Sciences - Food Science and Technology - Food Engineering
• Principal Investigator: Douglas Fernandes Barbin
• Grantee: Yasmin Lima Brasil
• Supervisor: Jose Manuel Amigo Rubio
• Host Institution: Faculdade de Engenharia de Alimentos (FEA). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Institution abroad: Universidad del País Vasco, Bizkaia (UPV), Spain
• Associated to the scholarship: 22/05540-2 - Study of starch-based materials using portable NIR spectroscopy and Hyperspectral Imaging (HSI) associated with artificial intelligence, BP.DR

Abstract

Brazil ranks fourth among the world's largest producers of plastic waste, with only 1.28% of this material being recycled, contributing significantly to environmental pollution. In this scenario, bioplastics made from cassava starch emerge as a promising alternative. These bioplastics offer antimicrobial and antioxidant properties when enriched with bioproducts such as oregano essential oil (OEO), encouraging the use of more sustainable materials. However, when considering the application of these polymers in fruit coating, it is crucial to assess the possibility of migration of film components to the surface of the food. This task is challenging due to the overlap of compounds present in smaller quantities, such as OEO. Advanced techniques, like near-infrared hyperspectral imaging (NIR-HSI) and X-ray fluorescence microscopy (XRF), enable the mapping of chemical compound distribution in foods. This analysis is facilitated by non-negative matrix factorization (NMF), which through clustering algorithms, simplifies complex data matrices into concentration profiles and pure spectral signatures. Thus, our goal is to evaluate the migration of OEO from films to the surface of strawberries using NIR hyperspectral images and XRF microscopy combined with machine learning techniques, in a fast and non-destructive manner, aligning with the requirements of Industry 4.0.