Metodología para la identificación de variables representativas en la captura de datos in situ de espectroscopía

Abstract

Spectral signatures are graphical representations that relate the electromagnetic spectrum to the wave response or amplitude, resulting from the interaction of objects and light. Each soil cover, due to the different composition of materials, generates a particular spectral response. Therefore, the automated classification of coverages, whether object-oriented or pixel-oriented, is based on the spectral response of the different segments or pixels, respectively. Spectral libraries are databases where these spectral responses are stored so that researchers or organizations involved in the collection, processing and analysis of these samples can work collaboratively, avoiding redundancy in the investigations and optimizing time in the development of the same. However, there is currently no defined metadata model to ensure interoperability between researchers or organisations that generate spectral signatures, so as a first step, it is necessary to identify those variables that turn out to be representative in the capture of spectral signatures. To make the respective parameter definition, different field coverage signatures are captured and spectrality is compared from the Spectral Angle Mapper - SAM algorithm.