基于Frangi滤波器的织物疵点检测方法  

Fabric defect detection based on Frangi filter

作　　者：张缓缓[1,2] 李仁忠[1] 李鹏飞[1] 景军锋[1] 赵娟[1] 

机构地区：[1]西安工程大学电子信息学院,陕西西安710048 [2]西安理工大学自动化与信息工程学院,陕西西安710048

出　　处：《毛纺科技》2015年第6期14-18,共5页Wool Textile Journal

基　　金：国家自然科学基金(项目编号:61301276);中国纺织工业联合会科技指导性项目(项目编号:2013066);陕西省教育厅自然科学专项(项目编号:2013JK0655);西安工程大学大学生创新创业计划项目(项目编号:201303012)

摘　　要：提出了一种基于Frangi滤波器的织物疵点检测方法。该算法首先对样本织物图像进行均值下采样处理,淡化和消除背景纹理对织物疵点检测的影响,然后将均值下采样处理后的图像经Frangi滤波器进行滤波,从而增强织物疵点部分以利于疵点的分割,最后对Frangi滤波后的图像进行阈值分割,分割出织物疵点部分。采用该算法对6种纹理织物进行处理,检测出26种疵点,92%的疵点能被准确的检测和定位,误报率为8%,检测效果较好。A new method based on Frangi filter was proposed to solve the problem of the fabric defect detection. Firstly, to eliminate the effect of the fabric texture, the fabric image was processed using mean sampling method, and then the processed image was filtered by Frangi filter to enhance fabric defect parts, finally the obtained filtered image was segmented and the fabric defect parts was detected. In this paper, six kinds of texture were detected and 26 fabric defects were found. In addition, 92% of the defects can be accurate detection and positioning, and false alarm rate was 8% , the results showed that the detection was better.

关 键 词：疵点检测 织物疵点 Frangi滤波器 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]