机器视觉在化工过程液位检测中的应用研究  

Application Research of Machine Vision in Liquid Level Detection in Chemical Process

作　　者：孟磊[1] 邹志云[1] 郭宇晴[1] 赵丹丹[1] 刘兴红[1] 

机构地区：[1]防化研究院,北京102205

出　　处：《石油化工自动化》2015年第4期39-43,共5页Automation in Petro-chemical Industry

摘　　要：液位检测在化工生产中有着广泛的需求,传统液位测量方式存在安装空间要求高、智能化程度低、组网难等问题。基于机器视觉技术设计了一套液位/界面测量系统:硬件上采用集成TCP/IP协议和接口的高速智能工业相机代替以往研究中CCD相机和图像采集卡,减少了系统空间占用,软件上利用图像预处理、Otsu阈值化和Harris角点检测识别液位,最后结合图像标定获取待测液位。实验表明:基于机器视觉的液位检测方法具有良好的精度,能够满足化工过程对液位检测的要求。Liquid level detection is commonly needed in chemical production. Traditional liquid measurement mode owns disadvantages of high requirement for installation space, low intelligence degree and difficulty of building up network. Based on technology of machine vision, a set of liquid level/interface detection system is designed. Intelligent industrial camera integrated with TCP/IP protocol is adopted instead of CCD camera and image capturing card to capture images with reduction of space in hardware. Image pre-processing, Otsu＇s threshold algorithm and Harris corner detection are utilized to identify liquid level in software. The liquid level will be achieved combined with camera calibration. Experiment results show the precision of the proposed method is high. It can satisfy need of level detection in chemical process.

关 键 词：机器视觉 液位检测 Otsu阈值化 HARRIS角点 

分 类 号：TQ02[化学工程] TP391.41[自动化与计算机技术—计算机应用技术]