设计了基于机器视觉的智能大输液异物在线检测系统,该系统可以实现灌装后大输液瓶中杂质的检测;研究了该检测系统的基本构成.工作原理以及一些关键图像处理技术;该检测系统采用特殊的硬件检测平台,让杂质与瓶身作相对运动,再利用精密的光学成像系统获取运动异物的序列图像,最后通过图像预处理、运动目标提取、运动目标分割和图像识别与判断这四个步骤完成对异物的检测;实验表明该系统可快速的检测出大输液溶液中所含的多种杂质,准确度高,实时性强。
The on--line detection system for foreign substances in a transfusion bottle of medicine solution was applied in the process of transfusion bottles production. The basic configurations, work principle and some critical image processing of the system were researched. The inspection system employ special hardware inspection platform, which made foreign substances move relatively to the bottle. And then it acquired a group of consecutive images with the help of delicate optic--imaging system. Finally, foreign substances were detected by image processing. This system can detect various foreign substances mixed in transfusion successfully and rapidly.