Investigating the 3D distribution of Cercospora leaf spot disease in sugar beet through fusion methods

Abstract

Cercospora leaf spot (CLS) disease, triggered by the fungus Cercospora beticola, represents the most severe foliar disease affecting sugar beets globally. The significant vertical heterogeneity of the plant canopy makes traditional 2D spectral imaging insufficient to accurately determining the CLS disease ratio. Integrating 3D and spectral imaging from dual sensors to form a plant spectral point cloud faces challenges due to alignment issues and high costs. An approach combining multi-view spectral images with the Structure from Motion (SfM) algorithm was introduced to generate a detailed multispectral 3D point cloud of plant structure. This technique was employed to assess the CLS disease ratio and its spatial heterogeneity. Specifically, a discriminant-based model was developed to differentiate healthy and diseased leaves using various ratio-based or normalized vegetation indices at the leaf scale. This model was then utilized to extract 3D CLS point clouds from the multispectral point clouds reconstructed by the new method at both plant and plot levels. Three-dimensional spatial heterogeneity analysis explored the vertical and horizontal distribution patterns of CLS in sugar beets. The findings revealed that disease levels determined by the 3D CLS model surpassed those of expert visual assessments (75 % vs. 58.3 %). The estimated disease ratio closely matched the measured values (RMSE = 8.4 %). Additionally, plot-scale CLS distribution maps aligned well with RGB image distributions. The analysis indicated that CLS initially spread from lower leaves upwards and displayed a pattern moving from the periphery to the interior. The introduced method offers a cost-effective, convenient alternative for generating detailed multispectral 3D point clouds. This study emphasizes the potential of spectral point clouds in monitoring plant canopy health and physiological activities.

Plant phenotyping relevance

マルチビュー分光画像とSfMを融合して3D植物点群を構築し、葉の病害状態と病害率を推定する方法が研究の中心であり、技術精度も評価している。

Code and data availability

公開状態または取得可能な本文経路を確認できませんでした。

No evidence-backed public reproduction asset is currently recorded.