欢迎来到天天文库
浏览记录
ID:55730932
大小:909.69 KB
页数:7页
时间:2020-06-03
《窗口加权协方差矩阵特征值检测织物瑕疵.pdf》由会员上传分享,免费在线阅读,更多相关内容在应用文档-天天文库。
1、第28卷第8期电子测量与仪器学报^289.82014年8月loURNALOFELECTRONICMEASUREMENTANDINSTRUMENTATION·885·DOI:10.13382/j.jemi.2014.08.012窗口加权协方差矩阵特征值检测织物瑕疵术康雪娟杨盼盼景军锋2(1.西安航空学院电气学院西安710077;2.西安工程大学电子信息学院西安710048)摘要:针对织物瑕疵严重影响布匹质量的问题,提出织物窗口加权协方差特征值自动检测织物瑕疵的方法。首先,选择织物窗口,以每个织物灰度值为中心的窗口遍历整幅织物图像,计算织物遍历窗口的加权协方差矩阵,从而
2、得出织物窗El加权协方差矩阵的特征值。织物瑕疵对应的加权协方差矩阵的特征值小于织物无瑕疵区域对应的特征值。织物无瑕疵织物窗口加权协方差特征值的均值作为织物瑕疵分割阈值。在检测阶段,根据分割阈值对待测织物的窗口加权协方差特征值进行判别,从而达到准确检测织物瑕疵的目的。实验结果表明,织物窗口尺寸范围在(11,31)时,织物的窗口加权协方差矩阵特征值可以准确快速检测织物瑕疵。避免了PCA多维特征值分类的过程,提高了检测织物瑕疵的效率,满足了织物工业在线检测需求。关键词:瑕疵检测;加权协方差矩阵;遍历窗口;特征值中图分类号:TP394.1;TN9文献标识码:A国家标准学科分
3、类代码:510.5015FabricdefectsdetectionbasedoncharacteristicvalueofwindowweightedcovariancematrixKangXuejuanYangPanpanJingJunfeng2(1.ElectricalEngineeringDepartment,Xi’anAeronauticalUniversity,Xi’an710077,China;2.CollegeofElectronicandInformation.Xi,肌PolytechnicUniversity,Xi’an710048,China)
4、Abstract:Aimingattheproblemofdefectinfluencingfabricqualitybadly,amethodtodetectfabricdefectispro-posed.Theproposedmethodcandetectfabricdefectautomaticallyusingthecharacteristicvalueofwindowweigh—tedcovariancematrix.Firstly,movingfabricwindowisselectedandmovingwindowbasedonthecenterofe
5、achfabricgreyvaluetraversesthewholeimage.Weightedcovariancematrixofmovingwindowonfabricsiscalculated,andtheweightedeigenvaluesofcovariancematrixareobtained.Theeigenvaluesofweightedcovariancematrixindefectsareaoffabricsarelessthantheabsenceofdefectareaoncorrespondingcharacteristicvalue.
6、Thresholdoffabricsegmentationistheaverageoftheeigenvaluesofweightedcovariancematrix.Indetectionsection,theeigen—valuesofcovariancematrixofdefectivefabricswouldbecalculatedtojudgetheexistenceofdefectbasedonthresh—oldoffabricsegmentation,thusaccuratedetectionoffabricdefectscouldbeachieve
7、d.Theexperimentalresultsshowthattheproposedmethodcandetectthefabricdefectsaccuratelyandspeedily,whenthesizeoffabricwindowiSselectedwithin(11,31).Atthesametime,muhi—dimensionalcharacteristicvalueclassificationofPCAcallbeavoidedandtheefficiencyoffabricdefectdetectioncanbeimproved.There
此文档下载收益归作者所有