Análisis de imágenes multiespectrales adquiridas con vehículos aéreos no tripulados

Abstract

This research paper deals with the experiences of the Group of Automation, Robotics and Perception (GARP) of the Universidad Central "Marta Abreu" de Las Villas (UCLV) and the Territorial Research Station of the Sugarcane (ETICA), in the use of an Unmanned Aerial Vehicles (UAV) for multispectral photogrammetry, with the aim at mapping experimental sugarcane plantations. Experimental tests with two of the best image processing software available at the market were done, Agisoft Photoscan and Pix4D, generating 3D orthomosaics and index maps NDVI (Normalized Difference Vegetation Index). Both programs were evaluated in terms of ease of use, processing time, and visual quality of the product, having the flight conditions of the UAV defined for an adequate processing of the images collected with the Parrot Sequoia Multispectral Camera. The results indicated a Pix4D program as the most efficient one to calculate vegetative indexes due to the incorporation of radiometric corrections specific for the orthomosaics. The results achieved were analyzed in collaboration with expert agronomists from ETICA, to make comparisons between the different groups of processed images and the correlation between the index maps with the variety of sugarcane crops, their age, the levels of plant vigor and other agricultural variables. The obtained results show that the spectral response permits to discriminate the fields depending on the age and demographic density in each parcel, but the NDVI don’t facilitate the determination of different varieties with the same plantation date