Detection of chestnut defect based on data fusion of near-infrared spectroscopy and machine vision

In order to improve the detecting precision of the eligible and defected chestnut, a multisource information fusion technique based on near-infrared spectroscopy and machine vision was proposed. Chestnuts from Jingshan area in Hubei province were taken as test samples. BPNN models for discriminating chestnuts based on near infrared spectroscopy, machine vision and multisource information fusion technique were built respectively. The discriminating rates of 3 models were 96.25%, 96.67% and 97.92% in training set, and 86.25%, 83.75% and 90.00% in prediction set. The overall results showed that it is feasible to discriminate chestnut quality using multisource information fusion technique. The recognizing rate of integration model was higher than that of the model built by machine vision technology or near-infrared spectroscopy alone.