基于机器视觉的钢棒自动计数系统  

Automatic counting system of steel rods based on machine vision

作　　者：成鹏飞[1] 杨帆[2] 言立强[2] 张马林[2] 王仲[3] 

机构地区：[1]长沙矿山研究院高压水射流技术研究所,湖南长沙410012 [2]湘潭钢铁集团有限公司 [3]装甲兵工程学院

出　　处：《冶金自动化》2008年第1期22-26,共5页Metallurgical Industry Automation

摘　　要：研究了一种基于机器视觉和图像处理的钢棒在线自动计数系统,为该系统设计了一种可从复杂背景图像中提取圆形目标的双阈值结合数学形态学运算的分割算法,并对如何在生产过程中获得复杂背景下钢棒的清晰图像,提取钢棒不同规格的特征参数进行了讨论。系统采用模块化硬件设计,图像处理软件满足实时检测的要求,可以有效地检测出打包的棒材数量。该系统在湘钢使用后,棒材识别率达98.8%,错误率为0,具有显著的经济效益和社会效益。An automatic on-line counting system of steel rods based on machine vision and image pro- cessing was studied. For this system, a method based on double-threshold segmenting combined with mathematical morphology was designed. Circular target can be extracted from complex background picture by use of the designed method. It is discussed that how to obtain clear picture of steel rods under complex background picture in production process and to extract characteristic parameters of steel rods with different specification. The system was designed with modular hardware. Image processing software can meet the need of reahime inspection and count steel rods effectively. After the system was put into application in Xiangtan Iron and Steel Group Co Ltd, steel rods recognition rate was over 98.8%, and error rate was zero. So, it has remarkable economic benefits and social effect.

关 键 词：图像识别 钢棒计数 机器视觉 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]