Detection of sclerotinia rot disease on celery using hyperspectral data and partial least squares regression

Abstract

There is a need to detect and assess the incidence of Sclerotinia rot disease in celery (Apium graveolens). In this study, we examined the potential of hyperspectral sensing to detect the symptoms of this disease in celery crop. Using a portable spectrometer, sample measurements of diseased and healthy leaves were collected from celery leaves in the field. Both raw and transformed spectral data were used in the development of Partial Least Squares regression models. The cross-validated results showed that the incidence of disease on celery could be predicted using the raw spectra and the first and second derivative data, with prediction errors ranging from 11.08 to 13.62%. The visible and near-infrared wavelengths (400-1300nm) produced similar detection ability with that of the full range wavelengths (400-2500nm).