Skip to Main content Skip to Navigation

Potentiel des images multispectrales acquises par drone dans la détection des zones infectées par la flavescence dorée de la vigne

Abstract : This work investigates the potential of remote sensing as a tool for the automatic detection of Flavescence dorée (FD) grapevine disease. The approach is based on the analysis of variables (spectral bands, vegetation indices, and biophysical parameters) computed from high resolution (10 cm) multispectral images and acquired by Unmanned Aerial Vehicle (UAV) during the period of maximum expression of symptom. The analysis of the variables discrimination performance is evaluated by a supervised method based on the Receiver Operating Characteristic curve (ROC curve). The training and validation areas used in this study were acquired from 14 vineyards located in southern France. The performance of the variables was tested on three different scales of analysis (one by plot, by cultivar and by berry color). Two levels of analysis have been implemented. The first level involves the potential of variables to discriminate Flavescence dorée symptomatic vines areas from asymptomatic ones. The second level of analysis is related to test the performance of the variables for the specific discrimination of Flavescence dorée vines (for the red cultivars) and the discrimination from Grapevine Trunk Diseases (GTD). The results obtained showed (1) a lower discrimination performance for discrimination of FD symptomatic vines areas from GTD symptomatic ones, more pronounced on the color level; (2) the presence of misclassified mixed pixels especially in the edges of the rows of vines and (3) a low discrimination of symptomatic vines areas (FD or MB) with a low proportion of symptomatic foliage (level of infection). From a thematic point of view, the results obtained showed the differences in the intensity of leaf discoloration affected by Flavescence dorée by year and their link with the chlorophylls and anthocyanins content of the leaves. Future prospects for this work concern the creation of a specific Flavescence dorée index depending on the color of the cultivars (red or white) and the intensity of leaf discoloration (attenuated or marked), identified from the hyperspectral data and improving the masking of mixed pixels from complex algorithms that consider the spatial distribution of pixels in the vine foliage.
Document type :
Complete list of metadatas

Cited literature [295 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Monday, September 30, 2019 - 3:01:07 PM
Last modification on : Saturday, January 9, 2021 - 3:31:12 AM
Long-term archiving on: : Monday, February 10, 2020 - 4:34:33 AM


Version validated by the jury (STAR)


  • HAL Id : tel-02301516, version 1



Johanna Leslie Albetis de la Cruz. Potentiel des images multispectrales acquises par drone dans la détection des zones infectées par la flavescence dorée de la vigne. Sciences de la Terre. Université Paul Sabatier - Toulouse III, 2018. Français. ⟨NNT : 2018TOU30157⟩. ⟨tel-02301516⟩



Record views


Files downloads