节点文献
基于支持向量机的射线检测焊接图像中缺陷识别
Defects Recognition Based on Support Vector Machine within Radiographic Testing Weld
【摘要】 提出了应用支持向量机(SVM)进行射线检测焊接缺陷识别的方法。该方法首先对图像进行预处理,并根据缺陷特点提取、选择8个参数作为特征参数,将焊缝内常见缺陷分为6类,根据有限的学习样本,建立影响缺陷类别的条件、因素和类别之间的一种非线性映射,对测试的样本进行识别。
【Abstract】 According to the problem of defect recognition within X-ray inspection weld,the method using SVM to recognize weld defects is put forword.This method preprocesses images firstly,and extracts and chooses 8 parameters as feature parameters according to the characteristics of defects,which are divided into 6 classes.Using the limited learning samples,a non-linear map is built up between the defect classes and the conditions and factors affecting them,and then testing samples can be recognized based on this map.
【关键词】 支持向量机;
多类分类;
焊接缺陷;
识别;
射线检测;
【Key words】 SVM; multi-class classification; weld defect; recognition; X-ray inspection;
【Key words】 SVM; multi-class classification; weld defect; recognition; X-ray inspection;
【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)
- 【文献出处】 煤矿机械 ,Coal Mine Machinery , 编辑部邮箱 ,2006年05期
- 【分类号】TP391.41
- 【被引频次】13
- 【下载频次】227