Monitoring of Crop Plant Height Based on DSM Data Obtained by Small Unmanned Vehicle Considering the Difference of Plant Shapes

Abstract

Unmanned Aerial Vehicle (UAV) has been recognized as a potential tool to obtain spatial information of crop state in small-scale farming fields. The objectives of this study are to assess the availability of UAV photographs to determine the plant height (PH) for three crop varieties (barley, oat, and perilla); and to evaluate the effect on the accuracy of the estimated PH caused by different plant shapes. The UAV photography and ground surveys were conducted weekly from May 31 to September 3, 2019, at the examination field of Obihiro University of Agriculture and Veterinary Medicine. The estimated PH was obtained from the differential value (D_value) of Digital Surface Models (DSMs) generated from UAV photographs through the Structure from Motion (SfM) process. The results demonstrate that the D_value of DSMs produced the highest estimation accuracy (R² = 0.97, RMSE = 9 cm) for perilla, which is a herb crop with luxuriant spade-shaped leaves, and a moderate estimation accuracy (R ²= 0.87, RMSE = 23 cm) for oat, which is a grain crop making small spikes during the heading time with lanceolate leaves. The estimated PH produced the lowest accuracy (R ²= 0.34, RMSE = 28 cm) for barley, which makes outward spikes with long prickly awns above the ear of grain. However, a higher accuracy (R² = 0.67, RMSE = 17 cm) for barley was obtained after a lodging of the spikes caused by rain on July 2. Furthermore, the accuracy of the estimated PH increased in both barley and perilla fields after corrected by plant coverage of the field. These findings suggest the D_value of DSMs obtained from UAV photographs can provide an accurate PH estimation for crops with luxuriant leaves like dicotyledons, but comparably less accurate estimation for crops with long and sharp leaves like monocots. Additionally, the elimination of the effect of unrecognizable factors such as long awns and non-plant factors such as the ground can also increase the accuracy of the estimated PH.