小型饮料瓶盖颜色识别系统设计与实现  

Design and implementation of small color recognition system for beverage bottle caps

作　　者：邓三星 孙赛伟 杨蒙蒙 张凯 李帅 DENG San-xing;SUN Sai-wei;YANG Meng-meng;ZHANG Kai;LI Shuai(Huanghe Jiaotong University,Jiaozuo,Henan 454950,China;Henan Intelligent Manufacturing Technology and Equipment Engineering Technology Research Center,Jiaozuo,Henan 454950,China)

机构地区：[1]黄河交通学院,河南焦作454950 [2]河南省智能制造技术与装备工程技术研究中心,河南焦作454950

出　　处：《食品与机械》2022年第4期121-126,共6页Food and Machinery

基　　金：河南省大学生创新创业项目(编号:S202113498008);黄河交通学院校级一流课程项目(编号:HHJTXY-2021ylkc58);黄河交通学院校级一流专业项目(编号:HHJTXY-2021ylzy01)。

摘　　要：目的:解决当前饮料瓶盖检测系统功能单一、体积偏大、颜色识别率低的问题。方法:提出一种基于ARM处理器的小型饮料瓶盖颜色识别系统设计方案,通过仿真软件HyperLynx的LineSim工具设计四层PCB板,设置传输线参数并进行阻抗匹配仿真分析;利用编译软件Jupyter Notebook实现阈值设置、目标轮廓检测、目标框出等功能。结果:在强、弱光条件下,识别系统对红、绿、蓝3种颜色瓶盖的识别率达到92.7%。结论:与傅里叶描述子相比,该识别系统识别准确率和精度更高,同时系统也具有人脸识别功能,适用于各种智能应用场景。Objective:This study focus on solving the problems that the current beverage bottle cap detection system has simple functions,large volume and low color recognition rate.Methods:A design scheme of small beverage bottle cap color recognition system based on ARM processor was proposed.By using the LineSim tool of the simulation software HyperLynx,a four-layer PCB board was designed and impedance matching simulation analysis was obtained after setting transmission line parameters.Through using the compilation software Jupyter Notebook to achieve threshold setting,target contour detection,target frame out and other functions.Results:Under bright and low light conditions,the bottle cap targets of red,green and blue were tested,and the recognition rate of the design scheme for the three colors could reached more than 92.7%.Conclusion:Compared with the Fourier descriptor,the scheme has higher recognition accuracy and precision,and the system also has the function of face recognition,which is suitable for various intelligent application scenarios.

关 键 词：机器视觉 瓶盖检测 阻抗匹配 颜色识别 轮廓检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TS27[自动化与计算机技术—计算机科学与技术]