光纤缺陷实时检测与分类方法研究

刘小勇; 郑琨

doi:10.7652/xjtuxb201409001

您当前的位置：

首页 >

文章列表页 >

光纤缺陷实时检测与分类方法研究

更新时间：2025-07-09

- 光纤缺陷实时检测与分类方法研究
- Research on Optical Fiber Defect Real-Time Detection and Classification
- 西安交通大学学报 2014年48卷第9期页码：1-5+18
- 作者机构：
  
  西安交通大学电子与信息工程学院,西安,710049
- 作者简介：
- 基金信息：
- DOI：10.7652/xjtuxb201409001
  中图分类号： TP274
- 网络首发：2014-09-10，
  
  纸质出版：2014
- 稿件说明：
移动端阅览
刘小勇, 郑琨. 光纤缺陷实时检测与分类方法研究[J]. 西安交通大学学报, 2014,48(9):1-5+18.

Research on Optical Fiber Defect Real-Time Detection and Classification[J]. 2014, 48(9): 1-5+18.
刘小勇, 郑琨. 光纤缺陷实时检测与分类方法研究[J]. 西安交通大学学报, 2014,48(9):1-5+18. DOI： 10.7652/xjtuxb201409001.

Research on Optical Fiber Defect Real-Time Detection and Classification[J]. 2014, 48(9): 1-5+18. DOI： 10.7652/xjtuxb201409001.

摘要

针对人工检测高速运动光纤表面缺陷的效率低、准确性差、难以实时检测等问题

设计实现了一套基于机器视觉的快速光纤缺陷检测系统。定义光纤缺陷

建立分类数据库和分类标准; 设计全方位数据采集系统自动连续获取光纤表面图像信息

输入工控机进行处理; 提取光纤目标区域

获得光纤缺陷形态学特征数据; 针对光纤缺陷特点和AdaBoost分类器的优缺点

设计了一种改进的基于形态特征的AdaBoost级联分类器用于光纤缺陷检测与分类

实现了光纤质量的实时监控。最后

将改进算法与标准AdaBoost算法在实际工业环境下进行对照实验

实验数据表明

改进算法的准确率达到99%以上

同时能大幅减少检测耗时

证明了所设计的检测系统具有很好的实时性和准确性。

Abstract

A real-time optical fiber defect detection system based on machine vision is designed to solve the problem that manual method is inefficient and imprecise in detecting fast moving fiber. Fiber defects are defined to establish a database and criterions for classification. Continuous fiber images captured automatically by 3 industrial cameras are transmitted to IPC for classification

and the IPC obtains morphological characteristics data of defects in fiber target after image processing. An advanced AdaBoost cascade classifier based on morphological characteristics is designed to analyze the defect image features. The advanced AdaBoost method and the classical AdaBoost method are tested under industry condition

and detection results show that the detection accuracy of the system with advanced AdaBoost is over 99% but the system with advanced AdaBoost takes much less time. It can be concluded that the proposed detection method has good real-time performance and a high detecting accuracy.

关键词

Keywords

references

贡玉南, 华建兴, 黄秀宝. 基于匹配Gabor滤波器的规则纹理缺陷检测方法[J]. 中国图象图形学报, 2001, 6(7): 14-18.

GONG Yunan, HUA Jianxing, HUANG Xiubao. Regular texture defect detection based on matched Gabor filters[J]. Journal of Image and Graphics, 2001, 6(7): 14-18.

胡涛. 基于轮廓对比的PCB裸板缺陷检测算法研究[D]. 武汉: 华中科技大学, 2009.

OTSU N. A threshold selection method from gray-level histograms[J]. Automatica, 1975, 11(1): 23-27.

FREUND Y, SCHAPIRE R E. A decision-theoretic generalization of on-line learning and an application to boosting[J]. Journal of Computer and System Sciences, 1997, 55(1): 119-139.

LIENHART R, KURANOV A, PISAREVSKY V. Empirical analysis of detection cascades of boosted classifiers for rapid object detection[C]∥Proceedings of the 25th German Pattern Recognition Symposium. Berlin, Germany: Springer, 2003: 297-304.

SCHAPIRE R E, SINGER Y. Improved boosting algorithms using confidence-rated predictions[J]. Machine Learning, 1999, 37(3): 297-336.

PAUL V, MICHAEL J. Robust real-time object detection, TR CRL 2001-01[R]. Cambridge, UK: Cambridge Research Laboratory, 2001.

VIOLA P, JONES M. Rapid object detection using a boosted cascade of simple features[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ, USA: IEEE, 2001: 511-518.

MEYNET J. Fast face detection using AdaBoost pattern recognition[EB/OL].[2003-07-16]. http:∥ftp.utcluj.ro/pub/users/nedevschi/AV/References/B3_Face%20detection/Meynet 2003_923.pdf.

赵锋, 薛惠锋, 王伟, 等. 采用多级分类器的手写数字识别技术研究[J]. 微电子学与计算机, 2006, 30(12): 30-32.

ZHAO Feng, XUE Huifeng, WANG Wei, et al. Handwritten digit recognition technology based on multistage classifying device[J]. Microelectronics Computer, 2006, 30(12): 30-32.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

采用试验设计方法的第三代CO2气体箔轴承箔片几何参数敏感性分析

语义区分和子图整合驱动的知识图谱表示学习方法

面向高海拔空冷发电机防晕性能演化规律的电热耦合研究

结构-语义-残差协同的图像压缩框架：面向隐私保护与高保真重建

离子液体改性燃料的电流体动力学效应研究