Predicción del color y humedad en granos de café variedad arábica (coffea arábica l.) usando imágenes Hiperespectrales (prediction of color and humidity in coffee beans arabic variety (coffea arabica l.) using hyperpephic imaging)

• Autor: Carranza Cabrera, Jhan Leymer
• Formato: tesis de grado
• Fecha de Publicación: 2017
• Institución: Universidad Nacional de Trujillo
• Repositorio: UNITRU-Tesis
• Lenguaje: español
• OAI Identifier: oai:dspace.unitru.edu.pe:20.500.14414/10037
• Enlace del recurso: https://hdl.handle.net/20.500.14414/10037
• Nivel de acceso: acceso abierto
• Materia: Hipercubo, imagen hiperespectral, color, contenido de humedad, regresión por mínimos cuadrados parciales

This study aimed to develop models to predict the color and moisture content of coffee beans cherry harvest state using hyperspectral imaging technique reflectance. Images were acquired from 82 grains cherry coffee arabica variety in ripening fruit, ripe and overripe (50 grains for model calibration and 32 grains for prediction) collected from Andrea, Shigua and Toribio plots in the province Toribio Rodriguez de Mendoza, Amazonas state; the spectral region used is between 400 and 1000 nm. The calibration models were constructed using regression by partial least squares (PLSR, acronym in English) obtaining a coefficient of determination (R2) of 0.95 and root mean square error of cross validation (RMSECV) of 3.30 for the a* value. Subsequently able to predict the color parameter a * cherry coffee beans arabica variety, obtaining a R2 of 0.87 and a root mean square prediction error of 2.0. The most influential wavelengths in prediction were selected using the β coefficients of the full spectrum models PLSR. The models obtained for moisture content and the L * and b * parameters were not suitable. PLSR simplified models were constructed using only selected wavelengths to predict the quality parameters, but the models didn’t have an acceptable level of prediction (R2 <70%).