Multispectral Imaging Supervised by Optical Spectrometry for Close Acquisition in Precision Agriculture

Satellite and airborne multispectral imaging has shown huge potential for many agricultural applications, primarily for plant indices. Imaging close to plants allows further expansion of this potential, such as early identification of plant disease or sufficient maturity status for harvest. To this aim, the joint use of optical spectrometry and multispectral imaging is shown effective: the most informative spectral bands for a given application can be identified so that the number of filters could be reduced to be mounted on a dedicated multispectral imaging system. To validate the on-field approach, we developed a compact, modular multispectral imaging system equipped with four sensors in the visible and near- IR band, a sensor in the far-IR, and a set of sensors to measure environmental parameters (T, RH, CO2, lux). This system collects only the most informative data needed for specific tasks from which spectral indices or processing methods can be implemented to improve automated analysis also through artificial intelligence approaches.