Quality Assessment of Jasmine Tea Combining Computer Vision and Color Difference Analysis

Jasmine tea faces challenges in quality assessment due to the short blooming period of jasmine flowers. This study investigates quality assessment method of jasmine tea based on the computer vision and color difference technology were conducted. Results showed that the free amino acids, polyphenols, and caffeine, exhibit significant correlations with sensory evaluation. Notably, the cultivar Dragon Tip Jasmine exhibited superior sensory attributes. The tea polyphenols content correlated highly significantly with a* (red-green component) (r = 0.69**) and ∆E (total color difference) (r = -0.78**). Caffeine content exhibits a highly significant correlation with I (brightness) value (r = 0.64**), a significant correlation with a* (r = 0.50*) and ∆E (r = -0.57*). A standard backpropagation (BP) neural network and a genetic algorithm-optimized backpropagation (GA-BP) neural network were constructed. The BP and GA-BP model of polyphenols parameters are as follows: input layer, a* and ΔE, output layer, the polyphenol content. The BP and GA-BP model construction of caffeine parameters are as follows: input layer, a*, ΔE, and the I value, output layer, the caffeine content. The GA-BP model demonstrated higher accuracy than the BP model for quality assessment. This research provides a novel and efficient approach for jasmine tea quality assessment.