Improving sugar beet canopy mapping through UAV image analysis.

Abstract

Fractional vegetation cover (FVC) is an important indicator of crop growth and a key parameter in vegetation modeling. Unmanned aerial vehicles (UAVs) equipped with RGB cameras offer a practical and cost-effective alternative to labor-intensive field surveys and multispectral imaging for FVC estimation. In this study, 18 segmentation methods, derived from the combination of six vegetation indices and three thresholding algorithms, were applied to UAV imagery of sugar beet fields during the 2022 growing season. The methods were validated using ground truth data collected from 30 plots across four growth stages. Results indicated that the Excess Green (ExG), Green Leaf Index (GLI), and Red-Green-Blue Vegetation Index (RGBVI), when combined with Otsu and Ridler-Calvard (RC) thresholding, generally provided the most accurate segmentation of vegetation cover. In particular, ExG with Otsu and RC achieved the highest accuracy (NRMSE = 5.1%, R² = 0.96), whereas ExGB with the Two-Peaks method showed the weakest performance (NRMSE = 42.3%, R² = 0.34). Statistical analyses confirmed that ExG-based approaches demonstrated stronger correlations with field measurements compared to other methods. These findings suggest that ExG in combination with Otsu or RC can be considered a promising option for UAV-based estimation of sugar beet vegetation cover, although further validation under different environmental and crop conditions is recommended.

Plant phenotyping relevance

UAV画像の植物領域分割手法を開発・比較し、地上真値で検証してサトウダイコンの植生被覆率を推定しており、表現型取得手法が研究の中心です。

Code and data availability

The paper's UAV sugar beet imagery, ground truth data, and analysis outputs are not publicly deposited; the authors state they are available only upon reasonable request from the corresponding author. No public code, dataset, or model repository is provided.

No evidence-backed public reproduction asset is currently recorded.