Bar Section Image Enhancement and Positioning Method in On-Line Steel Bar Counting and Automatic Separating System
Bar Section Image Enhancement and Positioning Method in On-Line Steel Bar Counting and Automatic Separating System
复制标题
DOI：
10.1109/cisp.2008.664
复制
发表时间：
2008-05
期刊：
影响因子：
--
通讯作者：
D. Zhang;Zhi Xie;C. Wang
中科院分区：
文献类型：
--
作者：
D. Zhang;Zhi Xie;C. Wang
在棒材厂的分离检验工段，需要对钢筋按规定的数量进行计数和分离，然后在下一道工序中进行打包。目前，国内外大多数钢筋厂仍采用人工清点钢筋，已不能满足自动化生产的要求。本文提出了一种基于计算机视觉的钢筋非线性计数与自动分选系统。该系统给出通过盘点区的钢筋数量，并通过驱动分条机将棒材分拣到一起。此外，本文还讨论了可信检测中的几个关键问题，如截面氧化、夹层和人体扰动等。通过模板匹配和可变阈值分割，提高了钢筋识别率的可靠性。在棒材厂的长期使用，证明了该方法的可靠性和实用性，误检率降至0.01个百分点以下。
At the separating and inspecting section of steel bar plant, steel bars have to be counted and separated by given number, then to be baled in the next procedure. Nowadays, most plants at home and abroad still count steel bars manually, which can't meet the requirement of roboticized production. This paper proposes a non-line steel bar counting and automatic separating system based on computer vision. The system gives out the amount of steel bars moving through the counting area, and separates bars by driving separating machine to tacked bar. Further more, this paper discusses some key problems in credible detection such as section oxidation, inter-cover and body disturbance. By template matching and mutative threshold segmentation, we improve the dependability of steel bars' recognition rate. Long time used in steel bar plant, this method is proved to be authentic and applied, the rate of misdetection has been curtailed to lower than 0.01 percent.