Abstract
At present, ''eating well" is increasingly desired by people instead of merely ''being full". Rice provides the majority of daily caloric needs for half of the global human population. However, eating quality is difficult to objectively evaluate in rice breeding programs. This study was carried out to objectively quantify and predict eating quality in Geng rice. First, eating quality and its components were identified by trained panels. Analysis of variance and broad-sense heritability showed that variation among varieties was significant for all traits except hardness. Among them, viscosity, taste, and appearance were significantly correlated with eating quality. We established an image acquisition and processing system to quantify cooked rice appearance and optimized the process of measuring cooked rice viscosity with a texture analyzer. The results show that yellow areas of the images were significantly correlated with appearance, and adhesiveness was significantly correlated with viscosity. Based on these results, multiple regression analysis was used to predict eating quality: eating quality = 0.37 × adhesiveness - 0.71 × yellow area + 0.89 × taste - 0.34, R(2) = 0.85. The correlation coefficient between the predicted and actual values was 0.86. We anticipate that this predictive model will be useful in future breeding programs for high-eating-quality rice.