摘要
分析了支持向量机的分类原理,指出在各类别样本数目相差较悬殊时,SVM不能获得良好的分类能力。针对焊接缺陷分类,提出了加权SVM(WSVM)算法。测试结果表明,该算法在焊缝RT图像中缺陷的分类识别中,能提高小类别缺陷的的检测精度,具有较高的理论和应用价值。
The paper analyses the classifying principle of the support vector machine(SVM),indicates that the SVM could not bear the excellent classifying capacity while the number of each kind of the sample differ greatly.According to the defect classification of the welding seam,the paper puts forward the weighted SVM algorithm.The testing result shows that the algorithm could increase the detection accuracy of the small kind defect on identifying the classification of the defect presented on the RT photo of the welding seam,which has higher theoretical and applicable value.
出处
《矿山机械》
北大核心
2006年第8期104-106,共3页
Mining & Processing Equipment
基金
江苏省博士后科研基金资助课题(2004035)
中国矿业大学科技基金资助课题(2005B005)。
