VIs-image segmentation method for the estimation of agronomic traits in durum- and winter-wheat cultivars

Vegetation indices (VIs) acquired by unmanned aerial vehicle (UAV) images are mainly associated with yield, yield components and crop physiological responses. Monitoring spatial and temporal variability of yield crop traits by remote sensing technique is a basis for a correct adoption of precision farming. The study reports a VIs-image segmentation method for the prediction of agronomic yield traits of seventeen wheat cultivars. The VIs-image segmentation was based on the combination of three VIs images (NDVI, SAVI and OSAVI) to improve the VIs sensitivity of the detection of dynamic monitoring and classification of wheat yield and yield crop traits at flowering stage. As expected, yield differences were recorded between durum and winter wheat cultivars. A cluster identified by high-resolution UAV for determining vegetation indices can be a valid strategy although its effectiveness is closely linked to the cultivar component and, therefore, requires extensive verification, since the genetic component can affect the accuracy.