Journal of Agronomy and Field Innovations

Background: The structure of a crop canopy is an essential metric that demonstrates the health of the plant as well as its biomass production and yields in precision agriculture. Traditional methods of measuring the crop in the field are labor-intensive, have limited spatial reach, and are susceptible to the errors of an observer.
Objective: This study proposes a methodology based on LiDAR technology to conduct non-destructive, precise measurements of the parameters of a crop canopy.
Methods: The study used point clouds obtained through LiDAR mounted on a UAV, which were then cleaned using filtering, segmentation, and extracted canopy information, to acquire canopy height, width, volume, and Leaf Area Index (LAI). A regression-based ML model was created and trained using the features obtained through that extraction.
Conclusion: The devised methodology enabled the determination of canopy height with RMSE equal to 4.8 cm and coefficient of determination R² equal to 0.94. This result is better than methods based on RGB and multispectral imaging while processing speed was enough to perform work on a field scale.