基于最小风险贝叶斯分类器的茶叶茶梗分类

目前在茶叶实际生产加工过程中,茶叶茶梗分拣自动化技术还处于不成熟阶段,分拣机械的精确度和效率还不能达到预期目的,必须通过再次人工分拣过程,大大增加了时间和人力成本。针对数码相机采集到的茶叶、茶梗数字图像,经过预处理后提取出样本的颜色和形状特征,并利用多元高斯模型进行建模,通过最小风险贝叶斯分类器对其进行分类。实验证明基于最小风险的贝叶斯分类器的分类方法是可行的,并取得了良好的分类效果。

Classification of tea and stalk based on minimum risk Bayesian classifier

Currently,in the process of actual production and processing of tea,the technology of tea-leaf and tea-stalk automational sorting is still in their infancy,and the precision and efficiency of sorting machinery hardly can achieve the desired objective.So the time and manpower costs must be increased again through the prcocess of manual sorting.In this paper,the digital camera is used to collect numeric pictures of tea-leaf and tea-stalk,then the color and shape features of these samples are extracted after pretreatment,and model is built with the use of multi-Gaussian model.The minimum risk Bayes classifier model is used to separate tea-leaf from tea-stalk.Experiments show that the minimum risk-based Bayesian classifier is feasible,and can obtain good classification results.