Classification of Tea Fermentation Duration by Image Processing and Artificial Neural Network

  • Ting Yao, Pandeng Lei, Hui Liu, Zhuoting Gan


Tea is a popular product that is consumed every day by a large number of people. Ranged from planting, harvesting, fermenting, and packaging, it is a long and complicated production process. Among these stages, fermentation plays a critical role in determining the tea quality. A computerized system is here developed for automatically classifying the fermentation duration by making use of tea leaf surface visual characteristics and an artificial neural network. Tea sample images from different fermentation duration are collected. Features are extracted based on probability distributions constructed from image hue and saturation. This information is then fed to an artificial neural network to perform classification. To guarantee high accuracy, an iterative training process is employed where the neural network is repetitively trained until a user-specified criterion is reached. Experimental results showed that the developed simple and efficient neural network scheme is promising for use in tea production.