Predicción de calidad y etapa de madurez en mango Kent (Magnifera indica) usando imágenes hiperespectrales

• Autor: Colchado Rojas, Ada Andrea Joselina
• Formato: tesis de grado
• Fecha de Publicación: 2019
• Institución: Universidad Nacional de Trujillo
• Repositorio: UNITRU-Tesis
• Lenguaje: español
• OAI Identifier: oai:dspace.unitru.edu.pe:20.500.14414/14826
• Enlace del recurso: https://hdl.handle.net/20.500.14414/14826
• Nivel de acceso: acceso abierto
• Materia: análisis espectral; análisis de imágenes; mango Kent; regresión lineal múltiple; tecnología de imágenes; atributos de calidad.

In the search for non-destructive inspection forms in fruit quality, in recent years there have been increased studies on the use of hyperspectral images in their quality. The objective of this investigation was to evaluate the level of prediction of quality parameters and maturity stage of the “Kent” mango by a model obtained using hyperspectral images. 120 mangoes were stored at four different temperatures (10, 12.5, 15 and 17.5 °C) at which images of both sides were taken in the near infrared range (890 - 1710 nm), one per day, for 12 days and with two repetitions for each fruit. Parallel to the taking of images, quality parameters such as firmness, color (ΔE*), brix degrees and moisture content were determined in the same mangoes. To obtain the model, the quality attributes were correlated with 70% of the spectral data resulting from the images using the partial least squares regression method (PLS-R). With the remaining 30% of the spectral data, the model was evaluated using Linear Multiple Regression (MLR). As a result, based on the selection of optimal wavelengths, the correlation coefficients (R2) were 0.12, 0.36, 0.33 and 0.66 and the Mean Square Root values of the Standard Prediction Error were 8.19, 14.24, 2.15, and 2.18 to predict color, moisture content, firmness and brix degrees respectively.