Storage Quality Prediction of Green Onions by Hyperspectral Imaging

Green onions are important flavoring food with a limited shelf life. Moisture and chlorophyll content are two important parameters for the post-harvest quality assessment of green onions. The aim of this paper was to obtain moisture and chlorophyll distribution of green onion under different postharvest storage conditions by means of a hyperspectral imaging (HSI) technique. The HSI was used to obtain the reflectance spectral data for green onions at 431~962 nm band. The original spectrum was transformed by three pretreatment methods of convolutional smoothing (SG), multiple scattering correction (MSC), and standard normal variation (SNV) to convert the original spectrum accordingly, and established the prediction model of moisture and chlorophyll content respectively. After comparing the prediction accuracy of the model, the MSC was found to have the best noise reduction effect was selected as the final spectral pretreatment method. Then a competitive adaptive weighted sampling method was used to select 11 and 20 optimal wavelengths for moisture and chlorophyll content predictions, respectively. Based on the selected wavelengths, partial least squares regression and support vector machine regression algorithms were used to establish the prediction model for moisture and chlorophyll contents. The prediction models based on the optimal wavelengths for moisture and chlorophyll content yielded 0.9046 and 0.9143, respectively. Finally, distribution maps of the moisture and chlorophyll content of green onions under the different storage conditions were obtained. In summary, the hyperspectral imaging might be used to rapidly detect the distribution of moisture and chlorophyll in green onion. This study would provide a theoretical basis for the subsequent development of portable measuring instruments for moisture and chlorophyll distribution in fruits and vegetables.