Identificación de la adulteración en el procesamiento de hojuelas de avena con kiwicha mediante espectroscopia NIR portátil e imágenes hiperespectrales
No Thumbnail Available
Date
2025
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Nacional de Trujillo
Abstract
El estudio aborda la detección de adulterantes en hojuelas de avena con kiwicha mediante espectroscopía NIR portátil e imágenes hiperespectrales (HSI). Con el objetivo de desarrollar un método rápido y no destructivo, se prepararon muestras con concentraciones de adulteración del 0.5% al 40% y se analizaron tres lotes diferentes. Se utilizó un espectrómetro NIR portátil para recolectar espectros de reflectancia y un sistema HSI para obtener imágenes detalladas. Los datos se analizaron mediante las técnicas quimiométricas PCA, PLS-DA y PLS, lo que permitió desarrollar modelos para clasificar y predecir la adulteración. Este enfoque busca mejorar la autenticidad de los productos alimentarios y su impacto en la salud y economía. Los resultados demostraron que ambas técnicas son eficaces para detectar adulteración. La espectroscopía NIR mostró un alto rendimiento en la identificación de arrocillo y avena caducada, con sensibilidades del 100%, pero tuvo dificultades con muestras puras. Las imágenes hiperespectrales, aunque más sensibles a variaciones físicas, presentaron mayor dispersión. Ambas técnicas presentaron una alta precisión con coeficientes de determinación (R²) superiores a 0.93 en la mayoría de los lotes, lo que indica su efectividad para la discriminación química, además las imágenes hiperespectrales complementaron el análisis, proporcionando una segmentación espacial detallada de los adulterantes, aunque mostraron una mayor variabilidad en concentraciones bajas (<5%). En general, la combinación de espectroscopía NIR y HSI permitió identificar adulterantes con una precisión superior al 98% en algunos casos, resaltando su potencial para aplicaciones industriales. El estudio presenta un enfoque innovador para abordar el problema de la adulteración alimentaria. Los resultados obtenidos son prometedores y la implementación de estas técnicas en el control de calidad de la industria alimentaria es viable y altamente beneficiosa. Se recomienda el desarrollo de algoritmos más avanzados y su validación en entornos industriales para optimizar la detección de adulteración en tiempo real y mejorar los estándares de calidad en la producción de alimentos.
ABSTRACT The study addresses the detection of adulterants in oatmeal flakes with kiwicha using portable NIR spectroscopy and hyperspectral imaging (HSI). Aiming to develop a rapid and non-destructive method, samples with adulteration concentrations ranging from 0.5% to 40% were prepared and three different batches were analyzed. A portable NIR spectrometer was used to collect reflectance spectra, and an HSI system was employed to obtain detailed images. The data were analyzed using chemometric techniques such as Principal Component Analysis (PCA), PLS-DA, and Partial Least Squares (PLS) regression, which allowed the development of models to classify and predict adulteration. This approach seeks to enhance the authenticity of food products and their impact on health and the economy. The results demonstrated that both techniques are effective for detecting adulteration. NIR spectroscopy showed high performance in identifying ground rice and expired oats, with sensitivities of 100%, but faced difficulties with pure samples. Hyperspectral imaging, while more sensitive to physical variations, exhibited greater dispersion. Both techniques showed high precision with coefficients of determination (R²) exceeding 0.93 in most batches, indicating their effectiveness for chemical discrimination. Additionally, hyperspectral images complemented the analysis by providing detailed spatial segmentation of adulterants, although they showed greater variability at low concentrations (<5%). Overall, the combination of NIR spectroscopy and HSI allowed for the identification of adulterants with over 98% accuracy in some cases, highlighting their potential for industrial applications. The study presents an innovative approach to addressing the issue of food adulteration. The results obtained are promising, and the implementation of these techniques in quality control within the food industry is feasible and highly beneficial. It is recommended to develop more advanced algorithms and validate them in industrial settings to optimize real-time adulteration detection and improve quality standards in food production.
ABSTRACT The study addresses the detection of adulterants in oatmeal flakes with kiwicha using portable NIR spectroscopy and hyperspectral imaging (HSI). Aiming to develop a rapid and non-destructive method, samples with adulteration concentrations ranging from 0.5% to 40% were prepared and three different batches were analyzed. A portable NIR spectrometer was used to collect reflectance spectra, and an HSI system was employed to obtain detailed images. The data were analyzed using chemometric techniques such as Principal Component Analysis (PCA), PLS-DA, and Partial Least Squares (PLS) regression, which allowed the development of models to classify and predict adulteration. This approach seeks to enhance the authenticity of food products and their impact on health and the economy. The results demonstrated that both techniques are effective for detecting adulteration. NIR spectroscopy showed high performance in identifying ground rice and expired oats, with sensitivities of 100%, but faced difficulties with pure samples. Hyperspectral imaging, while more sensitive to physical variations, exhibited greater dispersion. Both techniques showed high precision with coefficients of determination (R²) exceeding 0.93 in most batches, indicating their effectiveness for chemical discrimination. Additionally, hyperspectral images complemented the analysis by providing detailed spatial segmentation of adulterants, although they showed greater variability at low concentrations (<5%). Overall, the combination of NIR spectroscopy and HSI allowed for the identification of adulterants with over 98% accuracy in some cases, highlighting their potential for industrial applications. The study presents an innovative approach to addressing the issue of food adulteration. The results obtained are promising, and the implementation of these techniques in quality control within the food industry is feasible and highly beneficial. It is recommended to develop more advanced algorithms and validate them in industrial settings to optimize real-time adulteration detection and improve quality standards in food production.
Description
Keywords
Adulteración, Hojuelas de avena, Imagenes hiperespectrales (HSI), Espectroscopia NIR portátil, PCA. PLS-DA. PLS