作　　者： ; ; ; ;

机构地区： 湖南大学电气与信息工程学院

出　　处： 《电子测量与仪器学报》 2010年第2期125-130,共6页

摘　　要： 对比现有医药大输液药品可见异物人工检测方法,在线视觉自动检测具有巨大优越性,为此,本文设计了一套用于医药瓶内药液异物检测的视觉系统.首先研究了检测系统的机械与电气控制结构,开发了面向高速高精度医药生产线的图像获取装置;然后,基于图像中药液内异物运动轨迹的连续性,利用序列图像提取出图像中的运动信息,在此基础上,使用改进的Mean shift跟踪算法实现了可见异物的检测识别;最后选用100ml葡萄糖大输液进行在线测试,检测系统分辨率达到了国家药典的检测要求,系统运行准确率近95%,在线自动化视觉检测方法能很好满足医药生产线的要求. Compared with traditional medicinal solution foreign substance inspection manually, on-line machine vision inspection method has tremendous superiority. Therefore an automation inspection system based on machine vision is presented. Firstly, the mechanical structure and electrical control system are designed, the image acquiring device for high speed and high precision medicine product line is developed. Secondly, according as substances successive moving, the inspection method which employ an improved mean shift tracking algorithm is proposed based on a sequence images for transfusion liquid. Thirdly, choose 100ml 10% dextrose transfusion mixed in glass bottle as object, a detection rate of 95% at least is experimentally achieved for China pharmacopoeia. Experiment results demonstrate the proposed system can satisfy the requirement of medicine product line.

关 键 词： 医用大输液 视觉检测 可见异物检测 序列图像

领　　域： [自动化与计算机技术] [自动化与计算机技术]