基于机器视觉的柑橘分级技术研究  

Research on the Classification Technology of Citrus Based on Machine Vision

作　　者：王旭[1,2] 赵志衡[1] 

机构地区：[1]怀化学院机械与光电物理学院,湖南怀化418008 [2]怀化学院武陵山片区生态农业智能控制技术湖南省重点实验室,湖南怀化418008

出　　处：《怀化学院学报》2016年第5期60-63,共4页Journal of Huaihua University

基　　金：湖南省教育厅项目(15C1092);武陵山片区生态农业智能控制技术湖南省重点实验室项目(ZNKZ2015-3);怀化学院大学生研究性学习和创新性实验计划项目(2015)

摘　　要：以麻阳柑橘为研究对象,设计了由工业相机、光源、双平面镜等构成的全表面图像采集试验箱.在图像预处理的基础上,完成了柑橘横径、果形、成熟度、缺陷的检测.由隶属函数和权重集,采用加权平均算子计算综合评判结果,以最大隶属度原则确定柑橘等级.试验结果表明,优等果、一等果、二等果和等外果的分级正确率分别为100%、88.9%、84.6%、100%,该方法可以为武陵山片区农业智能化提供一定的参考.In this paper,a visual detection system for full surface image was built,which is composed of an industrial camera,a light source and two plane mirrors. The feature of Mayang citrus,such as transverse diameter,shape,maturity and defects was extracted based on image processing. By the membership function and weight set,the comprehensive result was computed by using weighted average operators, finally, the citrus grading can be obtained on the base of max subordination principle. The experimental results showed that the classification accuracy for the high- class,the first-class,the second- class and substandard citrus is 100%,88. 9%,84. 6% and 100%,respectively. This method can provide certain reference for the intelligent agriculture in Wuling mountain area.

关 键 词：机器视觉 柑橘 分级 模糊 

分 类 号：O439[机械工程—光学工程]