Investigation of a method to estimate the culm length of rice based on aerial images using an unmanned aerial vehicle (UAV) equipped with a high-precision positioning system

Unmanned aerial vehicle (UAV)-based remote sensing is an efficient method for evaluating plant traits in agricultural fields. In a paddy field, aerial images for the structure from motion (SfM) process were taken with a UAV equipped with a real-time kinematic-global navigation satellite system (RTK-GNSS). To evaluate the UAV-SfM approach for the remote sensing of plant traits, two analyses were performed: first, we determined the accuracy of the 3D coordinates of ground control points (GCPs) estimated with the SfM point clouds. Subsequently, the correlation between plant heights and culm lengths, which were predicted with the SfM point clouds and measured manually in the field, was analyzed. The errors on estimating the 3D coordinates derived from the SfM point clouds generated from image sets taken diagonally (camera angle at −60 degrees) were smaller than that of nadir image sets (camera angle at −90 degrees). The correlation coefficients (r) between plant heights predicted with the UAV-SfM approach at −60 degrees camera angle without using GCPs and culm lengths measured manually were 0.897–0.924 at a flight height of 25 m, 0.903–0.922 at 50 m, and 0.881–0.900 at 75 m. Therefore, the culm length of rice could be estimated with a UAV-SfM approach using image sets taken at a diagonal camera angle.