基于图像处理技术和神经网络实现机采茶分级

发布时间：2018-05-20 01:37

本文选题：机采绿茶 + 分级　；参考：《茶叶科学》2017年02期

【摘要】：为解决名优绿茶采摘环节的瓶颈问题,提出对机采大宗绿茶进行分级的思路。现有绿茶机采设备采摘的鲜叶一般只能制作普通的大宗绿茶,鲜叶存在混杂、破碎率高和老梗叶等问题,本文基于Labview vision、图像处理技术和神经网络算法分析机采绿茶成品的凸包面积、凸包周长、长轴长度、短轴长度等形态特征并对样本进行分类,实现从机采大宗绿茶中分选出名优绿茶。其中样本的形态特征采用工业CCD摄像头获取;用户界面用Labview自定义开发设计,数据交互方便,开发周期短。茶叶样本试验结果表明:该方案机采绿茶成品的分级正确率可以稳定在90%以上。本研究为进一步研究机采茶分级设备提供了良好的理论基础。
[Abstract]:In order to solve the bottleneck problem in picking process of famous and excellent green tea, the idea of classifying large quantity green tea by machine was put forward. In general, the fresh leaves picked by the existing green tea machine can only be made into ordinary large green tea, and there are problems such as mixed fresh leaves, high crushing rate and old stem leaves, etc. Based on Labview vision, image processing technology and neural network algorithm, this paper analyzes the shape features of green tea products, such as the area of convex hull, the circumference of convex hull, the length of long axis, the length of short axis and so on, and classifies the samples. To realize the separation of famous and excellent green tea from mechanical green tea. The morphological features of the samples are obtained by the industrial CCD camera, and the user interface is developed with Labview custom design, which makes the data exchange convenient and the development period is short. The results of tea sample test showed that the classification accuracy rate of green tea product was over 90%. This study provides a good theoretical basis for further research on the classification equipment of tea picking machine.
【作者单位】：安徽农业大学工学院;
【基金】：科技部科技型中小企业技术创新基金项目(14C26213401694)
【分类号】：TP391.41;TP183

【相似文献】