纹理织物疵点窗口跳步形态学法检测

发布时间：2018-01-31 12:14

  本文关键词： 纹理织物 形态学 跳步法 疵点检测　出处：《纺织学报》2017年10期 　论文类型：期刊论文

【摘要】：针对纹理织物疵点自动检测时因生产速度快造成的织物抖动以及检测速度难以匹配问题,提出窗口跳步形态学法纹理织物疵点检测算法。使用该算法对图像进行窗口分割及预处理后,首先对纹理织物图像的纹理特征进行分析,然后设计形态学算子进行腐蚀操作,最后使用连通域分析来确定疵点大小及位置。仿真实验及工厂实际应用表明,该算法可有效克服工业生产中纹理织物抖动造成的图像明暗不均,可检测出纹理织物中存在的破洞、经纬疵点、污渍、断线、折痕和结头等各种疵点,而且检测速度明显优于快速傅里叶变换特征点算法以及传统形态学检测算法。实时检测速度超过80 m/min,疵点检测精度为0.1 mm,满足实际生产需求。
[Abstract]:In order to solve the problem of fabric jitter caused by fast production speed and the difficulty of matching detection speed in automatic detection of texture fabric defects. A new algorithm of texture fabric defect detection based on window step morphology is proposed. After the image is segmented and preprocessed, the texture features of texture fabric image are analyzed. Then the morphology operator is designed to operate the corrosion. Finally, the size and position of the defects are determined by the connected domain analysis. The simulation experiment and the practical application of the factory show that. The algorithm can effectively overcome the uneven light and dark caused by the jitter of texture fabric in industrial production, and can detect all kinds of defects such as holes, warp and weft defects, stains, broken threads, creases and knots in textured fabrics. And the detection speed is obviously better than the fast Fourier transform feature point algorithm and the traditional morphological detection algorithm. The real-time detection speed is more than 80 m / min and the defect detection accuracy is 0.1 mm. Meet actual production requirements.
【作者单位】： 河北工业大学电子信息工程学院;天津市电子材料与器件重点实验室;北京市安视中电科技有限公司;
【基金】：中国博士后基金项目(2014M561053) 教育部人文社会科学研究规划基金项目(15YJA630108) 河北省自然科学基金项目(F2013202254) 河北省研究生创新资助项目(CXZZSS207035)
【分类号】：TP391.41;TS101.97
【正文快照】： 当前实际工业生产环境下,应用于生产线的织物疵点检测算法大都针对较简单且不含纹理的图像[1],例如文献[2-3]分别提出了针对坯布和非织造布的疵点自动检测算法,但这些算法难以检测更为常见的纹理型织物疵点。目前,国内外针对纹理织物的研究主要分为频率域与空间域处理方法。1)，

本文编号：1479040