[1]李玉凡,丁嘉伟,杜浩翠.doi: 10.3969/j.issn.1001-3849.2026.06.013电泳涂装线镀锌板表面杂质YOLOv8实时检测算法[J].电镀与精饰,2026,(06):105-112.
　LI Yufan,DING Jiawei,DU Haocui.YOLOv8 real-time detection algorithm for surface impurities on galvanized sheet in electrophoretic coating line[J].Plating & Finishing,2026,(06):105-112.
点击复制

doi: 10.3969/j.issn.1001-3849.2026.06.013电泳涂装线镀锌板表面杂质YOLOv8实时检测算法()

《电镀与精饰》[ISSN:1001-3849/CN:12-1096/TG]

卷:
期数:
2026年06
页码:
105-112
栏目:
出版日期:
2026-06-30

文章信息/Info

Title:
YOLOv8 real-time detection algorithm for surface impurities on galvanized sheet in electrophoretic coating line
作者:
李玉凡1; 丁嘉伟1; 杜浩翠2
(1. 郑州工业应用技术学院 信息工程学院,河南 新郑 451100 ;2. 河南师范大学 计算机与信息技术学院,河南 新乡 453007)
Author(s):
LI Yufan1;  DING Jiawei1;  DU Haocui2
(1. School of Information Engineering, Zhengzhou University of Industrial Technology, Xinzheng 451100, China; 2. College of Computer and Information Technology, Henan Normal University, Xinxiang 453007, China)
关键词:
电泳涂装线; 镀锌板; 表面杂质检测; 图像分割
Keywords:
electrophoretic coating line;  galvanized sheet;  surface impurity detection;  image segmentation
分类号:
TQ153；TG178；TN911.73
文献标志码:
A
摘要:
在高速生产线环境下，镀锌板表面杂质形态多样、背景干扰复杂，且因缺乏有效的镀锌板表面区域分割手段，难以精准提取并区分微小杂质与背景的细微特征，导致杂质位置识别不精准、类型误判率高，严重影响电泳涂装生产工艺的稳定性及产品质量。为此，提出一种电泳涂装线镀锌板表面杂质YOLOv8实时检测算法。利用Otsu阈值算法自适应分割镀锌板表面区域，排除背景干扰，为后续处理确定准确检测范围；基于分割后的镀锌板区域构建YOLOv8目标检测模型，利用其多尺度特征融合能力精准识别微小杂质区域；结合YOLOv8输出的杂质区域特征，通过随机森林算法检测出各微小杂质区域的具体杂质类型，以此实现表面杂质位置以及类型的实时检测。实验结果表明:该算法的平均交并比为0.95，表明其对镀锌板表面区域的分割效果良好，能够准确识别出电泳涂装线镀锌板表面图像中的杂质区域，且对杂质类型的检测准确性较高。
Abstract:
In the high-speed production line environment, the surface impurities of galvanized sheet are diverse and the background interference is complex. Due to the lack of effective segmentation methods for the surface area of galvanized sheet, it is difficult to accurately extract and distinguish the subtle features of small impurities from the background, resulting in inaccurate identification of impurity positions and high type misjudgment rates, seriously affecting the stability of electrophoretic coating production process and product quality. Therefore, a real-time detection algorithm YOLOv8 for surface impurities on galvanized sheet in electrophoretic coating line is proposed. Using Otsu threshold algorithm to adaptively segment the surface area of galvanized sheet, eliminate background interference, and determine the accurate detection range for subsequent processing. Constructing a YOLOv8 object detection model based on segmented galvanized sheet regions, utilizing its multi-scale feature fusion capability to accurately identify small impurity areas . By combining the impurity region features output by YOLOv8, the specific impurity types of each small impurity region are detected through the random forest algorithm, thereby achieving real-time detection of surface impurity positions and types. The experimental results show that the average cross-union ratio of this algorithm is 0.95, indicating that it has a good segmentation effect on the surface area of galvanized sheets, can accurately identify the impurity area in the surface image of galvanized sheets on the electrophoretic coating line, and has a high detection accuracy for impurity types

参考文献/References:

[1].李小丽, 苏会, 曾欢, 等. 无磷转化技术在商用车阴极电泳涂装中的应用研究[J]. 材料保护, 2023, 56(2): 183-187.
[2].姚士聪, 郝玉林, 林兴明, 等. 热镀锌板表面轮廓参数对电泳漆膜耐石击腐蚀性能的影响[J]. 电镀与涂饰, 2023, 42(13): 45-51.
[3].张忠, 魏国亮, 张之江, 等. 基于改进YOLOv5的新材料地板表面缺陷检测研究[J]. 包装工程, 2023, 44(7): 196-203.
[4].刘孝保, 张嘉祥, 阴艳超, 等. 主从特征融合驱动的铝型材表面缺陷检测[J]. 计算机集成制造系统, 2023, 29(1): 192-199.
[5].NI Y F, ZI Z X, CHEN W, et al. Steel strip surface defect detection based on multiscale feature sensing and adaptive feature fusion[J]. Journal of Real-Time Image Processing, 2025, 22: 65.
[6].许超杰, 李烨, 周子斐. 基于遮挡图像多尺度重建的表面缺陷检测[J]. 光学技术, 2023, 49(1): 70-77.
[7].何宇新, 陈勇, 王友元. 基于贝叶斯学习的数字化图像多阈值分割方法[J]. 计算机仿真, 2024, 41(5): 226-230+269.
[8].徐兴宇, 钟羽中, 涂海燕, 等. 基于深度半监督学习的小样本金属工件表面缺陷分割[J]. 计算机应用研究, 2024, 41(8): 2540-2545.
[9].朱永平, 程博, 熊聪, 等. 多尺度YOLOv5算法检测锂离子电池表面缺陷[J]. 电池, 2025, 55(1): 71-77.
[10].曾连连, 常强, 钟细明, 等. 热镀锌表面超疏水膜层构筑及其耐腐蚀性能研究[J]. 材料保护, 2025, 58(1): 40-48.
[11].王春梅, 刘欢. YOLOv8-VSC:一种轻量级的带钢表面缺陷检测算法[J]. 计算机科学与探索, 2024, 18(1): 151-160.
[12].丘锐聪, 周海峰, 陈颖, 等. 融合可切换空洞卷积的上下文信息增强船舶目标检测算法[J]. 大连海事大学学报, 2023, 49(4): 116-125.
[13].毕修平, 陈实, 张乐飞. 轻量级图像超分辨率的蓝图可分离卷积Transformer网络[J]. 中国图象图形学报, 2024, 29(4): 875-889.
[14].王媛彬, 吴冰超. 基于自适应特征融合和注意力机制的变电设备红外图像识别[J]. 电子与信息学报, 2024, 46(9): 3749-3756.
[15].赵川翔. 590 MPa级低合金高强度热镀锌钢板表面亮点缺陷分析[J]. 金属热处理, 2022, 47(6): 270-273.
[16].沈顺群, 宫旻, 刘凯, 等. 基于迁移学习和红外热成像的金属氧化物避雷器表面污染状态检测方法[J]. 电瓷避雷器, 2024(1): 1-8.
[17].罗龙, 李岩, 石岩, 等. 基于雾计算和随机森林算法的变压器温度监测及故障预测方法研究[J]. 变压器, 2023, 60(5): 15-22.
[18].张思, 金浩, 杨云锋. 基于Bootstrap抽样的厚尾自回归过程结构变点检测[J]. 华中师范大学学报(自然科学版), 2023, 57(4): 483-493.
[19].马二清, 靳立, 杨峥, 等. 热镀锌双相钢钢板表面橘皮缺陷研究[J]. 表面技术, 2023, 52(1): 214-221.
[20].胡琴, 刘黎, 徐兴, 等. 基于随机森林算法的异常发热复合绝缘子分类模型[J]. 电网技术, 2024, 48(1): 416-425.

更新日期/Last Update: 2026-06-12