In situ detection of moisture content and gelatinization degree during rice processing using hyperspectral imaging

During the industrial production of cooked rice, soaking and cooking are two critical processes that directly determine the quality of cooked rice. In this study, hyperspectral imaging technology was used to analyze and model the moisture content and the gelatinization degree of japonica rice (Wuchang rice and Long grain rice) and indica rice (Thai fragrant rice) during processing. The results showed that by analyzing the hyperspectral images, 35 characteristic wavelengths were obtained, and predictive models were established with the reflectance values as independent variables, moisture content and gelatinization degree as dependent variables. The best predictive models were partial least squares regression models, and the correlation coefficients and root mean square errors for the two models were 0.9673 and 0.5300, 0.9534 and 4.5046, respectively. Visualization of the moisture content and the gelatinization degree during processing were obtained using the models and enhanced by pseudo color and logarithmic transformation. The research results can provide theoretical and data basis for the development of intelligent rice processing production lines and reference methods for relevant research on hyperspectral quantitative modeling.