【作者】 侯浩通；

【导师】 陈磊；

【作者基本信息】 北京化工大学 ， 控制工程（专业学位）， 2015， 硕士

【摘要】 输液袋组件是一种常用的医疗用具,它的质量关系着使用者的健康和生命安全。以往,对于输液袋组件的检测多采用人工检测,输液袋组件体积较小,缺陷较难识别。人工检测存在效率低,准确性差等缺点。近年来,机器视觉技术快速发展,为输液袋组件的检测提供了一种新的思路。本文以输液袋组件为检测对象,研究了输液袋组件缺陷检测方法。本文分析了输液袋组件检测系统的要求,完成了检测硬件系统的设计,根据检测系统的要求,完成了相机,镜头的选型,硬件系统的搭建。研究了图像预处理算法。根据组件图像特点和检测要求,采用分数阶微分算子完成图像增强,采用加权中值滤波进行图像去噪,根据实验结果,选取Canny算子完成了组件图像的边缘提取,研究了利用最小二乘拟合的方法进行组件圆心定位。缺陷检测算法是本文的核心内容,根据待检测组件缺陷的特征,将组件正面分成多个区域,对各个区域设计了适用的缺陷检测算法。为验证本文提出算法的有效性,选取一定的数量的组件进行了实验。实验结果表明,本文提出的算法能够有效地检测出缺陷。利用Visual C#和Halcon、Matlab完成了缺陷检测软件设计,并介绍了软件的工作流程。更多还原

【Abstract】 Infusion bag component is a kind of important medical article, its quality affects users’health and safety. In the past, artificial Inspection is the main method for detecting the quality of infusion bag component. But artificial inspection has disadvantages such as low efficiency and low accuracy. The machine vision industry has enjoyed a great growth rate in recent years. The thesis takes Infusion bag component as the research object, studies the defects methods of the surface appearance.The hardware system of machine vision is designed in this paper. It consists of light source, camera, lens and PC.Pre-processing algorithm of infusion bag component image is studied in this paper.Based on the feature of infusion bag component image. Fractional differentiation is used to develop an enhancement method for infusion bag component image. Weighted median filtering algorithm is used to remove noise of the infusion bag component image. Based on the effect of four kinds of edge Inspection operator, the canny operator is selected to extract sub pixel contour. A method based on least squares fitting is developed to detect the centre of the infusion bag component image.Inspection algorithm of infusion bag component is the core of this paper. Firstly, the infusion bag component image is divided into some areas. Inspection algorithm is developed for each area. The experimental results show that inspection algorithm developed in this paper is effective for defect inspection.The software is developed in Visual Studio platform with Halcon and Matlab image processing function library. And the paper makes a detailed description for the designing and working principle of the software system.更多还原