一种茶叶茶梗色选机图像快速分拣方法  

A Quick Way Apply to Color Sorter Image to Sort Tea-leaf and Tea-stalk

作　　者：陈笋[1] 张春燕[1] 

机构地区：[1]安徽大学数学科学学院,合肥230601

出　　处：《合肥学院学报（自然科学版）》2013年第4期36-41,共6页Journal of Hefei University :Natural Sciences

基　　金：国家自然科学基金项目(61073116/F020508)资助

摘　　要：在茶叶生产加工过程中,为实现利用机器视觉技术进行茶叶茶梗分拣自动化,对茶叶色选机图像进行了研究,提出了一种基于最小错误率贝叶斯决策的茶叶茶梗快速有效分拣方法.针对数码相机采集到的茶叶、茶梗数字图像模拟实际生产加工的色选机图像,经过预处理后,提出最小外接圆半径与最大内切圆半径比形状特征,利用该单个形状特征进行高斯建模,依据最小错误率贝叶斯分类器对图像中的茶叶图像和茶梗图像的类别判断,从而实现茶叶茶梗目标图像的快速分类.实验结果表明,该方法在色选机图像分类中是一种实用和成功的方法.In the process of production and processing of tea, in order to implement the automation of tea and stalk sorting using roaching vision technology, the Tea Color Sorter image is studied, this paper propose tea and stalk sorting method based on the minimum error Bayes decision and this method is fast and effective. We take use of digital camera to collect numeric pictures of tea-leaf and tea-stalk and simulate the color sorter image in the actual production and processing. After preprocessing, we raise the shape feature that the ratio of radius minimum circumcircle＇ s with maximum inscribed circle＇ s radius and build gauss model using the single shape feature. Then apply the minimum error rate Bayes classifier to seperate the image of tea-leaf from tea-stalk, in order to achieve rapid classification of tea -leaf and tea-stalk＇ s target image. The experimental results show that this method is a practical and successful in the Tea Color Sorter image classification.

关 键 词：茶叶色选机 图像处理 数学形态学 最小错误率 贝叶斯决策 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]