Aerial-RGB imagery based 3D canopy reconstruction and mapping of grapevines for precision management

Abstract

This study aimed at exploring suitability of aerial-RGB imagery to map canopy vigor variability for precision vineyard management decision support. Unmanned aerial system with RGB imaging capability was used to image modern vertical shoot position trained vineyard multiple times in 2020 and 2021 field season. The vineyard had surface as well as deep root zone irrigation treatments of different levels (i.e., 100, 80, 60, 40% of evapotranspiration, ET). A custom algorithm was developed to 3D reconstruct the individual vine canopy and extract volume using convex hull method. The algorithm was successful in estimating canopy volumes with pertinent data being highly correlated ($r = 0.64$) with ground reference volume measurements. The resulting spatial volume maps also successfully quantified variation in irrigation treatments. Overall, the proposed high throughput canopy mapping approach can help growers to better understand vine canopy vigor variability throughout the production season and aid in vineyard management.