Diseño de una metodología experimental para la captura de imágenes hiperespectrales y su uso en la determinación de propiedades ópticas en muestras de la industria agrícola y energética

Abstract

Traditionally, the quality of agricultural products has been evaluated by visual inspection processes and/or chemical methods. However, in many cases, these methods are inopportune and insufficient because morphological changes in foods are evident in advanced stages of their development, making it difficult to restore their quality. In addition, chemical analyses require long analysis times, are expensive and do not allow real-time results to be obtained. In this context, optical spectroscopy techniques are presented as a fast, reliable, economical and environmentally friendly alternative for quality assessment and detection of irregularities in foods. These methods allow simultaneous quantification of different physicochemical properties of a sample through its spectral signature. Coffee is a product of great interest due to its high demand worldwide, which presents different sensory modifications through its processing. Therefore, these techniques are especially useful to make predictions about its quality. In this work, a real-time colorimetric analysis methodology has been developed, tested and standardized using RGB images for the quantification of the physical quality of cherry coffee during its processing. To this end, the chromatic information obtained from RGB and hyperspectral images has been compared in order to validate the colorimetric analysis method and its accessibility from any conventional camera. All this has been achieved through the use of a segmentation and chromatic analysis algorithm developed in Python.