Spectral bandwidth effect on a Rhizopus stolonifer spores detector and its on-line behavior using red tomato fruits

摘要

A fungal disease due to Rhizopus stolonifer causes important postharvest losses during shipping and marketing of tomatoes. A conidia detector that can be used on-site, quick, and without liquids was proposed. The feasibility of using near infrared spectroscopy (NIR) for detecting Rhizopus stolonifer conidia on red tomatoes was studied acquiring spectra before and after inoculating tomatoes in the laboratory. Discriminant analysis was carried out with 5 nm wide spectral data, and R. stolonifer conidiawas detected with an accuracy of 78%. The detection accuracy according to the sensing bandwidth was evaluated using 1 and 2 nm width signatures of spore-free and infected tomatoes. The accuracy detection of infected tomatoes increased to 88.92% with 1 nm wide spectral measurements. An automatic conveyor belt was developed to sample the inoculated tomatoes on-line using 1 nm spectral bandwidths acquired with a computerized spectrometer. Four different concentrations were applied and a 92% detection accuracy was encountered for a concentration of 6.5 x 10~4 sporangiospores/mL.