Análise de atributos biométricos de plantas cultivadas por meio de sensor RGB embarcado em veículo aéreo não tripulado

Abstract

Geotechnologies provides several techniques and tools that enable the analysis of spatial and temporal variability, of soils and plants, thus consolidating the practice of Precision Agriculture. Unmanned Aerial Vehicles (UAVs) play an important role in high spatial resolution aerial mapping. Although aerophotogrammetric activities, through UAVs, has advanced in the agricultural sciences, the generation of applied and automated technical products has not progressed as quickly, thus, the need for countless research remains. In this sense, the aim was to develop methods of analysis of orthophotomosaic and digital elevation models (from a RGB sensor embedded in UAV), through automated routines, to evaluate the quality of eucalyptus forest plantations (Eucalyptus ssp.) and 'Tahiti' acid lime orchard - (Citrus latifolia Tanaka) in the formation stage. Thus, it was also intended to develop a workflow, containing a set of aerophotogrammetric parameters that can be used, quickly, easily and reliably, in the biometric assessments of these cultures. For this purpose, an UAV, multirotor type, DJI Phantom 4 Adv, was used, equipped with a 20 Megapixel RGB sensor. The computational routines were developed in open source environments, consisting of three processing phases: i) Estimation of the Biometric Attributes of the Plant; ii) Statistical Analysis and iii) Statistical Map Report (SMR). The routines showed high accuracy in estimating the biometric parameters of the plants (above 93%) and good processing speed. Also, they are able to identify and count the plants, estimate the average spacing between plants and between rows, identify and quantify the planting failures, planting density, delimit and calculate the projected canopy area, height and volume of each plant. All analyzes are presented in a simple and intuitive way through SMR. The results were sufficient to define a method of obtaining aerophotogrammetric data that can be used to estimate and evaluate the quality of the biometric attributes of the aforementioned plantations, with the flexibility to adapt to other perennial crops.