摘要
Inthispaper,wepresentatiredefectdetectionalgorithmbasedonsparserepresentation.Thedictionarylearnedfromreferenceimagescanefficientlyrepresentthetestimage.Astherepresentationcoefficientsofnormalimageshaveaspecificdistribution,thelocalfeaturecanbeestimatebycomparingrepresentationcoefficientdistribution.Meanwhile,acodinglengthisusedtomeasuretheglobalfeaturesofrepresentationcoefficients.Thetiredefectislocatedbyboththeselocalandglobalfeatures.Experimentalresultsdemonstratethattheproposedmethodcanaccuratelydetectandlocatethetiredefects.
出版日期
2013年04月14日(中国期刊网平台首次上网日期,不代表论文的发表时间)