6533b7d8fe1ef96bd1269a9d
RESEARCH PRODUCT
Protocol for the Definition of a Multi-Spectral Sensor for Specific Foliar Disease Detection: Case of “Flavescence Dorée”
J. C. SimonFrédéric CointaultH. Al-saddikA. Laybrossubject
0106 biological sciences[SDE] Environmental SciencesDisease detectionComputer science[SDV]Life Sciences [q-bio]Multispectral imageradiometric/geometric correctionsFeature selectionMulti spectral01 natural sciencesfeature selection[INFO.INFO-TS]Computer Science [cs]/Signal and Image Processing[SDV.BV]Life Sciences [q-bio]/Vegetal Biology[SDV.BV] Life Sciences [q-bio]/Vegetal Biologytexture analysisProtocol (science)Artificial neural networkbusiness.industrymultispectral sensorOutbreakPattern recognition04 agricultural and veterinary sciencesFlavescence Dorée3. Good health[SDV] Life Sciences [q-bio]Identification (information)classification[SDE]Environmental Sciences040103 agronomy & agriculture0401 agriculture forestry and fisheriesFlavescence doréeArtificial intelligencebusiness010606 plant biology & botanydescription
Flavescence Doree (FD) is a contagious and incurable grapevine disease that can be perceived on leaves. In order to contain its spread, the regulations obligate winegrowers to control each plant and to remove the suspected ones. Nevertheless, this monitoring is performed during the harvest and mobilizes many people during a strategic period for viticulture. To solve this problem, we aim to develop a Multi-Spectral (MS) imaging device ensuring an automated grapevine disease detection solution. If embedded on a UAV, the tool can provide disease outbreaks locations in a geographical information system allowing localized and direct treatment of infected vines. The high-resolution MS camera aims to allow the identification of potential FD occurrence, but the procedure can, more generally, be used to detect any type of foliar diseases on any type of vegetation.Our work consists on defining the spectral bands of the multispectral camera, responsible for identifying the desired symptoms of the disease. In fact, the FD diseased samples were selected after establishing a Polymerase Chain Reaction (PCR) confirmation test and then a feature selection technique was applied to identify the best subset of wavelengths capable of detecting FD samples. An example of a preliminary version of the MS sensor was also presented along with the geometric and radiometric required corrections. An image analysis based on texture and neural networks was also detailed for an enhanced disease classification.
year | journal | country | edition | language |
---|---|---|---|---|
2018-10-26 |