作者：项新建, 尤钦寅, 郑雨, 黄炳强

作者单位：浙江科技学院自动化与电气工程学院，浙江 杭州 310023

关键词：珐琅礼盘;缺陷检测;曲线拟合;图像融合

摘要：

针对高端珐琅礼盘对美观度要求高，人工目检对尺寸大于100 μm的表面可视杂质识别困难等问题，该文提出一种珐琅礼盘细小杂质视觉检测方法。将采集到的灰度图像分为中心区域与边缘区域两类，对边缘区域图像进行阈值分割、最小二乘曲线拟合，提取出感兴趣区域；使用指定截止频率的傅里叶变换抑制两类图像中的细小噪声，获得检测区域图像；根据杂质的形态学特点设计基于图像融合的迭代算法，并通过测量最小外界矩形长边长度实现尺寸提纯，提高杂质的检测精度。实验结果表明：该方法对含有细小杂质缺陷样本的检出率为91.43%，单幅样本的平均检测耗时为577.4 ms，目标杂质漏检率为3.58%，可满足实际使用需求。

Detection of fine impurities in enamel gift dish based on image fusion
XIANG Xinjian, YOU Qinyin, ZHENG Yu, HUANG Bingqiang
School of Automation and Electrical Engineering, Zhejiang University of Science and Technology, Hangzhou 310023, China
Abstract: Aiming at the problems that high-end enamel gift plates have high requirements for aesthetics, and manual visual inspection is difficult to identify visible impurities on the surface with a size larger than 100 μm, this paper proposes a visual detection method for fine impurities in enamel gift dish. The collected gray images are divided into central area and edge area, then perform threshold segmentation and least square curve fitting on the edge area image to extract the region of interest. Use the Fourier transform with the specified cutoff frequency to suppress the small noise in the two types of images, and the detection area image is obtained. Finally, according to the morphological characteristics of the impurities, an iterative algorithm based on image fusion is designed, and the target impurity is purified by measuring the long side length of the minimum external rectangle, which improves the detection accuracy of impurities. The experimental results show that the detection rate of the defect samples of the fine impurities on the surface of the enamel gift dish by the method is 91.43%, the average detection time of a single sample is 577.4 ms and the target impurity missed detection rate is 3.58%, which meets the actual use requirements.
Keywords: enamel gift dish;defect detection;curve fitting;image fusion
2024, 50(4):24-30  收稿日期: 2022-03-28;收到修改稿日期: 2022-06-04
基金项目: 浙江省重点研发计划项目(2018C01085)；浙江省大学生科技创新活动计划暨新苗人才计划（2020R415032）
作者简介: 项新建(1964-),男,浙江永康市人,教授,硕士,主要从事人工智能、机器人、物联网理论与技术研究。
参考文献
[1] 赵凤奎. 人眼的最小分辨视角[J]. 唐山师范学院学报, 2008(5): 67-69.
ZHAO F K. Minal visual angle of people eyes[J]. Journal of Tangshan Teachers College, 2008(5):67-69.
[2] 王占云. 基于深度学习的搪瓷生产在线统计系统研究开发[D]. 郑州: 郑州大学, 2018.
WANG Z Y. Research and development of online statistical system for enamel production based on deep learning[D]. Zhengzhou: Zhengzhou University, 2018.
[3] 李捷. 基于机器视觉的贴花陶瓷盘表面缺陷检测方法研究[D]. 西安: 西安理工大学, 2019.
LI J. Research on the surface detect detection method of decal ceramic dish based on machine vision[D]. Xi’an: Xi’an University of Technology, 2019.
[4] 孙继旋. 激光雕刻中陶瓷网纹缺陷的自动检测方法研究[J]. 激光杂志, 2020, 41(8): 196-200.
SUN J X. Research on automatic detection method of ceramic mesh defects in laser engraving[J]. Laser Journal, 2020, 41(8):196-200.
[5] 吴越, 杨延竹, 苏雪龙, 等. 基于Faster R-CNN的钢板表面缺陷检测方法[J]. 东华大学学报(自然科学版), 2021, 47(3): 84-89.
WU Y, YANG Y Z, SU X L, et al. Surface defect detection method of steel plate based on Faster R-CNN[J]. Journal of Donghua University (Natural Science), 2021, 47(3):84-89.
[6] SANTOS J G, JANDERSON F, CRISTIAN M, et al. Ceramic cracks segmentation with deep learning[J]. Applied Sciences, 2021, 11(13): 6017
[7] 涂德浴, 刘坤, 朱庆, 等. 基于机器视觉的钢管壁厚在线检测方法研究[J]. 计算机工程与应用, 2022, 58(16): 249-256.
TU D Y, LIU K, ZHU Q, et al. Research on online detection for wall thickness of steel pipe based on machine vision[J]. Computer Engineering  and Applications, 2022, 58(16):249-256.
[8] 项新建, 李超, 尤钦寅. 基于图像增强的气门弹簧座内壁划痕检测算法[J]. 中国测试, 2023, 49(5): 9-15.
XIANG X J, LI C, YOU Q Y. Valve spring seat’s scratches algorithm based on image enhancement[J]. China Measurement  & Test, 2023, 49(5):9-15.
[9] LIU Q M, LI X. Machine vision detection on circle with non-uniform points[J]. Applied Mechanics and Materials, 2014, 3634(687-691): 819-822
[10] 胡浩, 李俊峰, 沈军民. 基于机器视觉的小磁瓦表面微缺陷检测方法研究[J]. 机电工程, 2019, 36(2): 117-123.
HU H, LI J F, SHEN J M. Detection methods for surface micro defection on small magnetic title based on machine vision[J]. Journal of Mechanical & Electrical Engineering, 2019, 36(2):117-123.
[11] 赵丽梦. 基于图像融合的绝缘子缺陷检测方法研究[D]. 西安: 西安石油大学, 2021.
ZAO L M. Insulator defect detection based on image fusion[D]. Xi’an: Xi’an Shiyou University, 2021.
[12] 化春键, 邓朝省, 陈莹. 基于双掩模图像差影的工业产品表面缺陷检测[J]. 传感器与微系统, 2015, 34(5): 127-129.
HUA C J, DENG C S, CHEN Y. Industrial products surface defects detection based on image subtraction with double masks[J]. Transducer and Microsystem Technologies, 2015, 34(5):127-129.
[13] SHOLEHKERDAR A, TAVAKOLI A, LIU Z. Theoretical analysis of tsallis entropy-based quality measure for weighted averaging image fusion[J]. Information Fusion, 2020, 58(C): 69-81
[14] 曹义亲, 武丹, 黄晓生. 基于改进蚁群算法的轨道缺陷图像分类[J]. 计算机科学, 2019, 46(8): 292-297.
CAO Y Q, WU D, HUANG X S. Track defect image classification based on improved ant colony algorithm[J]. Computer Science, 2019, 46(8):292-297.