Aplicación de la tecnología de imágenes hiperespectrales al control de calidad de productos agroalimentarios de la región de Amazonas (Perú)

Abstract

[EN] The evaluation of the quality of food products is a constant concern for the different actors in the production chain and requires the use of a number of analytical methods: physical, chemical, biochemical and sensory; which are destructive, time-consuming, are expensive, laborious and, in the case of sensory, subjective, which makes its application process online. Unlike traditional methodologies, hyperspectral imaging technology emerges as a non-destructive tool for quality assessment. In the Amazon Region (Peru) quality evaluation is performed with traditional methodologies, which has significant limitations as, for example, late detection of infections in crops, if coffee rust; the inability to determine adulteration, if the presence of starch in cheese; or, the subjective classification as quality standards, case marbling of beef. On the other hand, the application of hyperspectral imaging technology requires a series of steps for obtaining, processing and analysing information, which have been developed and implemented by various researchers, but have not been unified in one system and procedures for use by users unfamiliar with the depth of the computer analysis. In this context, this thesis has been planned in order to develop an IT application for the use of hyperspectral imaging technology to assess the quality of three food products in the Amazon Region (Peru). For this purpose it was necessary to create a number of subsystems, modules and routines, which were subsequently implemented by user windows (Guides) and functions using the mathematical software Matlab 2010a. The applicability of the developed system was tested in three case studies: detecting the degree of rust infection in samples of coffee leaves detecting adulteration in samples of fresh cheese with starch at different concentrations, and classification according to samples marbling standards beef. As a result of the system development, the developed IT toolswere implemented in a software that allows easy use of them and that is compatible with the system by reflectance imaging. As for the cases studied, the analysis results of early infection coffee leaf rust, showed statistically significant differences from the analysis judges attributable to the increased sensitivity of the system. For the analysis of fresh cheese adulteration has been shown that the application of the hyperspectral imaging technique to detect adulteration cornmeal in cheese, based on the relationship between the reflectance and the starch content in samples of fresh cheese. Also, the implemented methodology, based on hyperspectral imaging technology, proved to be valid for the determination of the quality of beef based on the Japanese standard of marbling. An advanced version of the software could be patented and transferred to industry, at a later stage to the realization of this thesis. Such software could be designed so that could include progressively new useful tools in quality control of other agro-food processes.; [ES] La evaluacion de la calidad de los…