基于局部二值模式的茶青纹理分析及其应用研究

项目名称： 基于局部二值模式的茶青纹理分析及其应用研究

项目编号： No.31470028

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 农业科学

项目作者： 汤哲

作者单位： 中南大学

项目金额： 30万元

中文摘要： 由于目前的茶叶加工装备中缺乏有效检测茶叶状态的手段,即使先进茶叶生产线的茶叶加工过程还依赖人工干预,不能实现智能化加工。本项目通过茶叶加工技术和信息技术的交叉与深度融合，从新的角度提出实现茶叶智能化加工的方法。首先，本项目对茶青纹理智能分析在茶叶加工技术中的应用前景和理论展开深入的研究；其次，利用计算机视觉技术分析茶青状态，提出改进的局部二值模式结合灰度共生矩阵对茶青纹理特征进行提取，这样既能充分提取纹理特征又能满足茶叶连续化加工的实时性要求；最后，从茶叶加工工艺和智能技术着手，利用支持向量机对茶青纹理进行特征识别，并用神经网络等智能技术建立从茶青纹理特征到茶叶加工参数的映射，使生产线的数据采集与监控系统能根据茶青的类型和构成自动选择茶叶生产线的最佳加工参数，摆脱对人工的依赖，真正实现茶叶加工装备的智能化。通过本项目的研究，为茶叶加工装备从连续化到智能化的升级提供新的有效手段。

中文关键词： 茶叶加工；茶青纹理特征；灰度共生矩阵；支持向量机；数据采集与监控

英文摘要： Due to the lack of effective tools for the tea leaf status detection, manual operations are still necessary during the automatic tea process even for advanced tea production lines. As a result, it is difficult to achieve intelligent tea process. A new approach to the intelligentization of tea process machines has been proposed in this project by the fusion of tea process technologies and information technologies. First, to analyze the texture properties of fresh tea leaves by using computer technologies, an improved LBP (Local Binary Pattern) in combination with GLCM (Gray-Level Co-occurrence Matrices) has been proposed for tea leaves texture extraction. This method is able to effectively extract fresh tea leaves texture and meet the real-time operation requirements of continuous tea process. Second, by the integration of tea process and intelligent technologies, fresh tea leaves texture recognition has been achieved through SVM (Support Vector Machine). At the same time, a map between the fresh tea leaves texture properties and the tea process parameters has been created by using the intelligent control techniques (e.g. neural networks), which enable the SCADA (Supervisory Control and Data Acquisition) to select an optimal set of tea process parameters according to any kind of tea leaves type and rid the depen

英文关键词： tea processing；tea green texture characteristics；Gray Level Co - occurrence Matrix；Support Vector Machines；Data acquisition and monitoring