Assessment of a 3D canopy model using UAV-based multispectral imaging in precision viticulture

Abstract

Several studies have provided many techniques to rapidly characterize canopy architecture using the canopy height and width. Biomass is one of the most important parameters for the farmer as it can be useful to choose the best canopy management within the field and can be estimated using plant canopy height. Such information is processed and summarized in a 3D rows model identified as canopy height model (CHM). This study presents the development of an unmanned aerial vehicle based on a Mikrokopter Okto able to acquire multispectral images with high resolution and process them through a data processing workflow. The image processing system makes it possible to construct a 3D surface model (DSM) for the production of digital terrain models at the highest detail (DTM) using photogrammetric algorithms computing vision-based techniques Structure from Motion (SFM). The DTM was then subtracted from the DSM to obtain the canopy height model (CHM) of the vineyard. The results show a good separation of the soil from the pixel rows. A further comparison between CHM and vigour map based on the NDVI index showed a good correlation and a preliminary assessment of the biomass volume using the average height and the average width of the crown for three different homogeneous classes was made. This is a preliminary study about how a 3D model developed by high-resolution images captured by the Unmanned Aerial Vehicle (UAV) can be useful to make a quick and accurate assessment of the biomass.