Optimization strategy for black tea digital blending by fusing image and spectral information

Screening, picking and blending are the three major processing procedures for refining tea. Among them, blending is the key step in improving the appearance and flavor of tea, and can provide a standardized tea product in large quantities. Traditionally, tea blending is performed by experienced blending masters based on the sensory review results of each raw tea. In this study, machine vision is used to obtain golden pekoe, color and cord as the main image information. Near-infrared spectroscopy is used to analyze the various components in tea to evaluate the taste of tea. Then, the image information and spectral information are fed into the black tea digital blending model to realize the digital blending of black tea, and the results are verified via professional evaluation. The GDE3, NSGAII and NSGAIII were chosen to convert the tea blending problem into a multi-objective optimization problem considering both cost and quality. The convergence speed and degree of GDE3 algorithm are better than the other two algorithms. This study reveals that digital blending of black tea provides a novel and effective approach for tea standardization.