摘要
针对医疗针管刻度缺陷检测存在样本收集难且缺陷种类不确定的问题,提出一种少量缺陷样本情形下的医疗针管刻度质量检测方法;利用真实生产线上采集的大量正常样本,训练深度刻度分割模型,通过以针管像素分块为单位构建拉普拉斯矩阵来挖掘针管刻度间相关性,并利用模糊C均值进行无监督缺陷检测;实验结果表明,针管刻度质量检测方法可100%检测出所有缺陷样本,对提高医疗针管生产质量具有理论和实践应用价值。
Aimed at the problems of difficulty in sample collection,and uncertainty of defect types in medical syringe scale defects,a syringe scale quality detection method for limited defective samples is proposed.The large normal samples collected from the actual production line are used to train the deep scale segmentation model.The syringe pixels are divided into blocks as the unit,the Laplacian matrix is constructed to mine their correlation,and the fuzzy C-means is used to carry out the unsupervised defect detection.Experimental results demonstrate that this syringe scale quality inspection method can detect all defective samples,with a precision of 100%,effectively enhancing the production quality of medical syringes.
作者
严菲
黄海燕
谢致尧
王晓栋
YAN Fei;HUANG Haiyan;XIE Zhiyao;WANG Xiaodong(School of Computer and Information Engineering,Xiamen University of Technology,Xiamen 361024,China;School of Computer Engineering,Guangzhou City University of Technology,Guangzhou 510800,China)
出处
《计算机测量与控制》
2024年第7期70-76,共7页
Computer Measurement &Control
基金
国家自然基金项目(U1805264)
厦门市科技计划项目(3502Z20227073)
福建省自然科学基金项目(2021J011186,2023J011428)
福建省教育中青年项目(JAT200486)
福厦泉自创区协同专项(3502ZCQT2021009)
广州城市理工学院2022年度校级青年科研基金项目(K0222006)。
关键词
缺陷检测
模糊聚类
图像分割
深度学习
机器学习
defect detection
fuzzy clustering
image segmentation
deep learning
machine learning
