Challenges and Bottlenecks in VAV Phenotyping

摘要

Plant phenotyping using unmanned aerial vehicles (UAVs) needs to be implemented with appropriate experimental design and corresponding software tools to prevent bottlenecks in the processing chain. In this study 24 durum wheat varieties were compared at two locations in NW and central Spain with irrigated and rain-fed treatments, but at one trial location it was necessary to protect the crops using a plastic mesh. Apart from agronomic components, a field spectrometer and multispectral UAV sensor were employed. The MosaicTool software plugin was developed in the FIJI image analysis platform for quick, semi-automatic plot level extraction from multispectral UAV orthomosaics, but field to UAV spectral data correlations reduced from r=0.95-0.97 to 0.64-0.91 when the mesh was present. Linear Spectral Unmixing (LSU) with renormalization is proposed to correct the UAV data to remove the effects of the protective mesh.