基于图像处理的织物疵点检测算法综述  

Overview of Fabric Defect Detection Algorithms Based on Image Processing

作　　者：刘小敏[1] 刘国高[1] 宗国华[1] 张卓[1] 张学武[1] 刘宇兴 

机构地区：[1]河海大学物联网工程学院,常州213022 [2]常州世纪鸟电动车制造有限公司,常州213118

出　　处：《微处理机》2016年第1期37-42,共6页Microprocessors

基　　金：国家自然科学基金资助项目(61273170);常州市科技支撑项目(CE20140008)

摘　　要：织物表面疵点的检测在视觉检测领域应用十分广泛,对纺织品制造业中产品质量的监控具有重要的现实意义。目前织物疵点检测仍存在很多难题,基于检测精度、通用性和实时性这三个标准,对近年来织物疵点检测算法进行了对比和总结。检测算法根据检测对象广义上被分为两大类:无图案的表面疵点检测和有图案的表面疵点检测。前者使用一些传统的算法就能检测出疵点,而后者相对复杂,使用传统的算法很难完成检测。将目前提出的算法分为以下几类:统计法、频谱法、模型法和有图案背景的织物检测,对几种典型的算法进行深入比较,分析其优缺点,并总结了织物疵点检测仍面临的挑战以及未来的发展趋势。Fabric defect detection,widely used in the vision detection field,plays an essential role to ensure the quality of products in the textile manufacturing industry.At present,there are still a lot of problems in the detection methods.Such criteria as detection accuracy,versatility and instantaneity are utilized to assess and compare the detection algorithms in recent years,which can be broadly divided into pattern -based detection and non -pattern -based detection.Non -pattern -based approaches are conventional,while pattern -based approaches are relatively complex.The algorithms,including the statistical algorithm,spectral algorithm,model -based algorithm and pattern -based one,are presented to analyze its advantage and disadvantage,and make in -depth comparison between them.At last,a summary is made for the challenge on the detection technologies and the future research.

关 键 词：织物 疵点检测 检测精度 通用性 实时性 无图案 有图案 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]