Bibliographic citations
Magán, C., Vásquez, N. (2017). Modelado de propiedades texturales durante el proceso de maduración de queso tipo suizo utilizando perfiles espectrales [Tesis, Universidad Privada del Norte]. https://hdl.handle.net/11537/12361
Magán, C., Vásquez, N. Modelado de propiedades texturales durante el proceso de maduración de queso tipo suizo utilizando perfiles espectrales [Tesis]. PE: Universidad Privada del Norte; 2017. https://hdl.handle.net/11537/12361
@misc{renati/418395,
title = "Modelado de propiedades texturales durante el proceso de maduración de queso tipo suizo utilizando perfiles espectrales",
author = "Vásquez Medina, Nadya Paola",
publisher = "Universidad Privada del Norte",
year = "2017"
}
ABSTRACT Typically evaluating the texture during cheese maturation requires invasive and destructive methods as well as specialized equipment. In this work, we propose a non-destructive method based on hyperspectral images. The objective of this study was to perform modeling of the textural properties of Swiss-type cheese during the ripening process using their spectral profiles. Thirty-two samples were prepared and stored in ripening conditions: [0 - 5] weeks, 20°C and 80% humidity. Hyperspectral images were acquired in the range of 400 to 1000 nm, along with the ripening process, and conducted the profile texture analysis (PTA) using two compression cycles. In PTA were obtained hardness values (H) for each compression cycle (H1 and H2). Then, the hyperspectral images were pre-processed and analyzed to obtain the average spectral profiles using functions which were implemented in Matlab 2015a. Modeling and optimization of the relationship between spectral profiles and the hardness values were performed by the method of partial least squares regression (PLSR), obtaining the adjusted models by minimizing the predictive residual sum of squares (PRESS). The relationship modeling showed that the most relevant wavelength for H1 model were 592, 496, 576, 456, 512, 560, 680, and 624 nm; and for the H2 model were 592, 496, 576, 456, 512, 624, 680, and 560 nm (sorted by influence on the model). Finally, it was calculated the adjusted correlation coefficient (R2) obtaining values of 0.89 and 0.90 for H1 and H2 respectively. It is possible to predict textural values using a model based on spectral profiles with an accuracy near to 90% and the more relevant wavelength for PLSR model are in the range of 450 to 680 nm.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.