基于机器视觉的武夷岩茶做青程度识别研究

Study on the color degree recognition of Wuyi Rock Tea with machinery vision

[目的]尝试利用机器视觉技术分析武夷岩茶做青过程中萎凋叶的颜色变化情况和做青程度之间的关系。[方法]试验通过图像采集系统获得做青过程的茶叶彩色图像,采用RGB和HSI这2种颜色参数模型观察图像颜色变化,结合自组织竞争人工神经网络技术对不同程度的做青叶进行分类。[结果]140个样本以RGB分量值和HSI分量值表示的训练集和测试集分类的正确率分别为81.64%、78.6%和76.5%、73.8%,显著性检验概率P<0.05,决定系数R^2最小为0.879,颜色分量和做青程度高度相关。[结论]试验构建的自组织竞争人工神经网络模型可以作为武夷岩茶做青程度的预测模型,也可为今后利用机器视觉来监测乌龙茶做青过程提供参考。

[Objective] The objective of this study was to analyze the correlation between the color change of tea leaves after sun withered and the degree of green processing using machinery vision technology. [Methods] Color images were captured by an image acquisition system in green processing phase, and the color change of tea leaves was monitored by RGB and HSI models combined with self-organized artificial neural network technology to classify different degrees of leave green processing. [Results] The classification accuracy rates of training set and testing set of 140 samples expressed by RGB component and HSI component were 81.64%,78.6% and 76.5%,73.8%, respectively. The probability(P)of significance testwas less than 0.05, and the coefficient of determination R^2 was equal or greater than 0.879. The color component was highly correlated to the degree of green processing.[Conclusion] The self-organized artificial neural competition network model can be used to predict the degree of Wuyi Rock Tea green processing,and to provide a reference for monitoring the process of Oolong tea green processing by machinery vision in the future.