Estimation of Ground Surface and Accuracy Assessments of Growth Parameters for a Sweet Potato Community in Ridge Cultivation

Abstract

There are only a few studies that have been made on accuracy assessments of Leaf Area Index (LAI) and biomass estimation using three-dimensional (3D) models generated by structure from motion (SfM) image processing. In this study, sweet potato was grown with different amounts of nitrogen fertilization in ridge cultivation at an experimental farm. Three-dimensional dense point cloud models were constructed from a series of two-dimensional (2D) color images measured by a small unmanned aerial vehicle (UAV) paired with SfM image processing. Although it was in the early stage of cultivation, a complex ground surface model for ridge cultivation with vegetation was generated, and the uneven ground surface could be estimated with an accuracy of 1.4 cm. Furthermore, in order to accurately estimate growth parameters from the early growth to the harvest period, a 3D model was constructed using a root mean square error (RMSE) of 3.3 cm for plant height estimation. By using a color index, voxel models were generated and LAIs were estimated using a regression model with an RMSE accuracy of 0.123. Further, regression models were used to estimate above-ground and below-ground biomass, or tuberous root weights, based on estimated LAIs.

Code and data availability

The supplied blocks describe UAV-SfM phenotyping of a sweet potato field (point clouds, DTM/PHM, LAI voxel models) but contain no data availability statement, no public dataset or code deposit, and no author-provided URLs. Analysis used commercial software (Agisoft PhotoScan, IDL 8.0, ENVI) with no shared scripts or模型的

No evidence-backed public reproduction asset is currently recorded.