A Novel Hyperspectral Imaging Approach for Early Detection of Broomrape Infestation in Tomato Plants

Abstract

Abstract Broomrape (Orobanche spp.) is a root-parasitic weed that severely threatens tomato crops by siphoning nutrients during subterranean development, often causing irreversible yield losses before above-ground symptoms appear. This study investigates a non-invasive approach for early detection of broomrape in tomato by combining hyperspectral imaging with narrow-band vegetation indices. Tomato plants were imaged with a ground-based hyperspectral camera. We applied statistical and machine-learning models to distinguish infested from healthy plants at stages prior to broomrape emergence. We found that vegetation indices sensitive to chlorophyll and canopy vigor (e.g. NDVI, NDVIre, PSNDb, GNDVI) showed significant declines in infested plants, reflecting parasite-induced stress. Using these indices in a classifier yielded high accuracy in early infestation detection. Our results demonstrate that hyperspectral-derived indices can reveal the physiological impact of Orobanche parasitism on tomato leaves before visible symptoms. This approach offers a promising tool for precision agriculture, enabling targeted management of broomrape (e.g. site-specific herbicide application) and improving crop protection strategies.

Code and data availability

The preprint describes hyperspectral imaging of tomato plants to detect broomrape infestation, with vegetation index tables and K-means preprocessing. No public repository, dataset, image, or code deposit is provided; the authors state the underlying data are available only on request from the corresponding author.

No evidence-backed public reproduction asset is currently recorded.