Strawberry Anthracnose Disease Assessment Using Hyperspectral Imaging

Early detection of plant diseases can help preventlosses of agricultural production and the spread of diseases. Anthracnose disease is one of the most devastating diseases that threaten strawberries.The objective of this research is to study non-destructive methods for the early detection of strawberry anthracnose disease using hyperspectral imaging techniques. Standard protocols to acquire hyperspectral images were established for strawberry leaves inoculated with anthracnose disease. Experiments were performed, involving reflectance of hyperspectral images from 400~1000 nm, to observe infected strawberry leaves on a daily basis. The hyperspectral images enable the investigation of the temporal spectral changes and the spread of infected areas in the leaves. It was found that the spectra of disease in symptomless and symptomatic areas in leaves significantly differ at the wavelength ranges between 540~570 nm and 750~900 nm. The spectral data of the non-infected, infected, and symptomatic areas in leaves were further analyzed and classified with the multivariate analysis method (MVA). The results also revealed that the spread of the infected area in leaves can be predicted one day before it can be detectedvisually. The results from this research show that the hyperspectral imaging technique provides an efficient and novel tool for plant pathologists in strawberry anthracnose disease assessment. Furthermore, the developed methods will facilitate early detection of strawberry anthracnose disease in fields so that proper plant disease control measures or management can be executed.