Nondestructive detection of maturity of watermelon by spectral characteristic using NIR diffuse transmittance technique

This study demonstrated a spectral characteristic analysis method used in the maturity detection of watermelon, which was based on the visible and near-infrared spectroscopy (Vis/NIR) technology. We found that with the gradually physiological maturity of watermelon, for the diffuse transmittance spectra, the intensity of peak s (around 720-740 nm) decreased and peak(2) (around 802-805 nm) increased, and the intensity of peak s was always higher than peak(2) for immature watermelon, on the contrary, the intensity of peak) . was always lower than peak(2) for mature watermelon. We proposed two spectral characteristic analysis methods that one was using the ratio of the intensity between peak s and peak(2) (RPP), the other one was using normalized difference intensity of peak (NDIP) to determine maturity stage (immature, mid-mature, mature and over-mature). And a correction factor based on the boundary value is used in RPP and NDIP to improve the classification ability. The discriminate performances of these two methods were compared through the performance of least-squares support vector machines (LS-SVM). The result indicated the Corrected-RPP (C-RPP) method showed the best predictive ability with 88.1% of correct classification rate. This is a new finding that the maturation of watermelon can be tested directly from the spectra. It has a great potential to be developed to study other fruits and be applied in portable instrument and on-line high-speed watermelons internal quality detection system based on spectroscopy database.