基于多分辨率的全局及局部显著性的色织物疵点检测  

Yarn-dyed fabric defect detection based on multi-resolution global-local saliency

作　　者：韩士星 李鹏飞[1] 张银河 HAN Shixing;LI Pengfei;ZHANG Yinhe(School of Electronic and Information,Xi’an Polytechnic University,Xi’an 710048,China;2011 Shaanxi Province Technical Textiles Collaborative Innovational Center,Xi’an Polytechnic University,Xi’an 710048,China)

机构地区：[1]西安工程大学电子信息学院,陕西西安710048 [2]西安工程大学,陕西省2011产业用纺织品协同创新中心,陕西西安710048

出　　处：《纺织高校基础科学学报》2020年第2期23-29,共7页Basic Sciences Journal of Textile Universities

基　　金：国家自然科学基金(61301276);重点产业创新链(群)-工业领域项目(2019ZDLGY01-08)。

摘　　要：为提高色织物疵点检测效率和准确性,提出一种融合多分辨率的全局及局部显著图的色织物疵点检测方法。针对织物疵点及纹理的频谱特征,通过高斯金字塔生成不同分辨率图像,采用小波变换的全局显著性算法,以及傅里叶变换融合的方法得到全局综合显著图。使用基于图论的显著性(graph-based visual saliency,GBVS)算法计算织物图像局部显著性。加权融合全局及局部显著图得到综合显著图,并进行图像分割及形态学操作对疵点区域进行检测。使用不同算法对5种不同类型色织物疵点进行检测,结果表明:提出的方法的疵点检测率达92.0%,与传统方法相比有较大提高,且检测时间较短。In order to improve the accuracy and efficiency of yarn-dyed fabric defect detection,a method of yarn-dyed fabric defect detection based on multi-resolution global and local saliency map is proposed.According to the spectrum characteristics of fabric defects and textures,different resolution images are generated by Gaussian pyramid,and the global saliency algorithm of wavelet transform and the method of Fourier transform fusion are used to obtain the global comprehensive saliency image.Then the graph based visual saliency(GBVS)algorithm is used to calculate the local saliency of the fabric image.Finally,weighted fusion of global and local saliency map is used to get comprehensive saliency map,and image segmentation and morphological operation are carried out to detect defect areas.Through the defect detection of five different types of yarn-dyed fabric pattern,it shows that the defect detection rate of the method is 92.0%,which is much higher than the traditional method,and the detection time is shorter.

关 键 词：色织物 疵点检测 多分辨率全局显著性 综合显著图 图像融合 

分 类 号：TS101.9[轻工技术与工程—纺织工程]