基于纹理增强分水岭的织物瑕疵检测新算法  

A new method for fabric defect detection based on texture enhancement and watershed

作　　者：沈晶[1] 况晓静[1] 张量[1] 吴先良[1] 

机构地区：[1]合肥师范学院电子信息工程学院,安徽合肥230061

出　　处：《合肥工业大学学报（自然科学版）》2014年第7期812-817,共6页Journal of Hefei University of Technology：Natural Science

基　　金：国家自然科学基金资助项目(61301062;51207041)

摘　　要：由于受到织物表面附着有细小绒线、结头、灰尘等材质以及成像噪声的影响,织物图像的纹理特性在一定程度上被模糊,给织物缺陷检测带来了困难。针对上述问题,文章提出了一种基于纹理增强分水岭的织物瑕疵检测新算法。该算法首先利用非局部均值滤波实现纹理增强,消除成像噪声及材质的影响,有效地突出了纹理间的差异。在此基础上,进一步利用纹理分水岭方法,提取缺陷区域;通过对一组纺织品缺陷检测的实验结果表明,该算法能够准确地提取出纺织品的缺陷区域,实现纺织品缺陷检测。The texture property of fabric image is affected by the fine knitting yarn ,rolling and dust attached on the surface of fabric product as well as imagery noise ,which will cause the texture property of fabric image to be blurred to some extent .This makes the fabric defect detection more difficult .To solve this problem ,a new method for fabric defect detection based on texture enhancement and watershed is proposed in this paper .In this method ,the texture enhancement can be achieved by using the nonlocal means（NL-means） filter ,so the effect of imagery noise and materials is eliminated and the difference between the defect texture and background texture can be obviously enhanced .Combined with the watershed transform ,the defect regions in fabric images can be detected accurately .The experimental result show s that the proposed method achieves efficient and accurate performance on the detection of fabric images containing fabric defects .

关 键 词：图像分割 纺织品缺陷 纹理增强 小波变换 分水岭 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]