作者：李建军¹, 史志晖¹, 崔桂梅¹, 张帅¹, 王磊²

作者单位：1. 内蒙古科技大学信息工程学院，内蒙古 包头 014010;
2. 包头钢铁（集团）有限责任公司热轧部，内蒙古 包头 014010

关键词：机器视觉;边缘点;坐标提取;边缘矢量;镰刀弯检测

摘要：

针对热连轧车间里带钢的镰刀弯检测问题，提出一种基于机器视觉的镰刀弯在线检测方法。首先，对预处理后的图像采用图像二值化处理提取目标，再使用形态学运算消除目标上的杂质以及不连通区域。然后在图像中以图像的中心点像素作为原点建立坐标系，提取中间坯两侧边缘点，接着获取相邻点之间的向量，并进行单位绝对值化，计算相邻单位向量之间的关系，得出中间坯边缘的弯曲度，依此判决镰刀弯。选取以现场获取的视频所分割出的5000张带有镰刀弯的图像作为实验样本，在PyCharm和OpenCV平台上进行实验。实验结果表明，该方法的镰刀弯检测成功率达$ 96.32\mathrm{\%} $，能够更准确地检测带钢镰刀弯，解决运行非对称镰刀弯检测方法因带钢跑偏而引起的误检问题。

Research on sickle bend detection method based on edge vector algorithm
LI Jianjun¹, SHI Zhihui¹, CUI Guimei¹, ZHANG Shuai¹, WANG Lei²
1. School of Information Engineering, Inner Mongolia University of Science and Technology, Baotou 014010 ,China;
2. Hot Rolling Department , Baogang Group, Baotou 014010 ,China
Abstract: Aiming at the problem of scythe detection of strip steel in hot strip mill, an online scythe detection method based on machine vision was proposed. First, image binarization is used to extract the target from the preprocessed image, and then morphological operation is used to eliminate impurities and disconnected areas on the target. Then in the image with the center of the image pixel as the origin to establish a coordinate system, extract the edge points on both sides of the middle billet, then get the vector between adjacent points, and carry out unit absolute value, calculate the relationship between adjacent unit vectors, get the curvature of the middle billet edge, according to the decision of the sickle bend.

images with scythe curves segmented from videos obtained on site were selected as experimental samples, and experiments were conducted on PyCharm and OpenCV platforms. The experimental results show that the success rate of sickle bend detection is 96.32%, which can detect the sickle bend of strip steel more accurately, and solve the misdetection problem caused by strip deviation caused by the asymmetric sickle bend detection method.
Keywords: machine vision;edge point;coordinate extraction;edge vector;sickle bend detection
2022, 48(1):14-19  收稿日期: 2021-01-12;收到修改稿日期: 2021-03-05
基金项目: 国家自然科学基金资助项目（61763039，62066036）
作者简介: 李建军（1977-），男，内蒙古乌兰察布市人，副教授，博士，研究方向为目标检测与跟踪
参考文献
[1] КЛЮШИН Д, 张绍庆. 螺旋焊管生产过程中带钢镰刀弯的自动检测装置[J]. 重型机械, 1982(6): 71-72
[2] 邝家涛, 史志呈. 板坯镰刀弯检测系统在CSP生产线的开发和应用[J]. 南方金属, 2011(5): 45-47
[3] 刘桂雄, 张瑜, 蔡柳依婷. 机器视觉检测图像拼接融合技术研究进展[J]. 中国测试, 2020, 46(1): 1-6
[4] 张青春, 王旺, 杨广栋. 基于多目立体视觉的机械臂智能控制系统设计[J]. 中国测试, 2020, 46(12): 79-85
[5] 刘洋, 徐冬, 王晓晨, 等. 热连轧运行非对称测控系统研究与应用[J]. 冶金自动化, 2020, 44(1): 48-54
[6] CHEN Z Y, QI J J, LIU H Q, et al. Research on production technology of asymmetrically hot rolled stainless steel clad plate[J]. Materials Science Forum, 2020: 6053
[7] LOHUMI S, CHO B K, HONG S D. LCTF-based multispectral fluorescence imaging: System development and potential for real-time foreign object detection in fresh-cut vegetable processing[J]. Computers and Electronics in Agriculture, 2021: 180
[8] 曾晟, 梁乃兴, 薛轲, 等. 基于扩展极大值变换沥青混合料数字图像预处理方法研究[J]. 中外公路, 2020, 40(2): 225-228
[9] 吴昆鹏, 杨朝霖, 石杰. 基于钢板轮廓精确检测的智能优化剪切方法的应用[C]. 中国金属学会: 中国金属学会, 2019: 7.
[10] 叶兵, 马伟东. 可变算子图像边缘复杂目标特征识别方法仿真[J]. 计算机仿真, 2019, 36(10): 453-457
[11] SIKKA P, ASATI A R, SHEKHAR C. Real time FPGA implementation of a high speed and area optimized Harris corner detection algorithm[J]. Microprocessors and Microsystems, 2021: 80
[12] 李国荣, 刘玉芝. 基于视觉检测系统的建筑尺寸偏差标注方法研究[J]. 自动化与仪器仪表, 2020(9): 182-185
[13] 陈黎艳, 熊强强. 光图像亚像素边缘高精度自适应检测研究[J]. 激光杂志, 2020, 41(11): 86-90
[14] 师平, 沈宝国. 基于机器视觉的工件角度测量方法[J]. 装备制造技术, 2012(11): 127-128+131
[15] 刘勇, 杨竹, 何登贵, 等. 热连轧中间带坯镰刀弯检测判定分析系统开发[J]. 金属材料与冶金工程, 2020, 48(1): 42-47