Automatic sorting of fresh tea leaves using vision-based recognition method

Mechanical tea harvesting using plucking machines is highly efficient, but harvested raw fresh tea leaves (FTLs) are always low quality because they contain a mixture of old leaves and leaf debris. To address this problem, this study developed an automatic sorting machine with a vision-based recognition method to extract high-quality FTLs from plucked raw FTLs. First, the raw FTLs were separated one by one after passing through three sequential conveyor belts with increasing speed, and were then classified into four grades using a vision-based recognition method. Finally, the FTLs were blown by air nozzles into collection boxes according to their specific grade. In the recognition method, the shape-based feature of each FTL is extracted by establishing the FTL's topological structure, and the support vector machine model is used for classification. The experimental results revealed that the vision-based recognition method performed satisfactorily with an accuracy rate of 94% and precision rate of 85%. The sorting success rate and efficiency of the automatic sorting machine were approximately 80% and 15 kg hr(-1), respectively. The results indicate that the developed automatic sorting machine can effectively and efficiently sort raw FTLs, which may improve the profitability and promote the automation of tea processing.