基于支持向量机的X射线焊缝缺陷检测  

Defect Detection of X Ray Image of Weld Using Support Vector Machine

作　　者：陈方林[1] 刘彦[1] 

机构地区：[1]中北大学,山西太原030051

出　　处：《机械工程与自动化》2010年第2期122-123,126,共3页Mechanical Engineering & Automation

基　　金：国家自然科学基金资助项目(60602041)

摘　　要：为提高对焊缝缺陷的检测精度,提出采用支持向量机(Support Vector Machine,SVM)分类的方法对X射线焊缝图像进行分割。选择训练样本图像的灰度、形态学梯度作为训练向量的特征分量对SVM进行训练,得到SVM分割模型后,将测试样本输入分割模型进行分割处理。以气孔缺陷为例,证明了该方法能实现焊缝气孔缺陷的准确分割,与其他分割方法相比,可提高缺陷检测的精度。In order to improve the detection precise of weld, an image segmentation method based on Support Vector Machine （SVM） used for X ray image of weld was proposed. The image gray-scale and morphological gradient were used as training vector to train the SVM, after the SVM segmentation model was obtained, the test samples would be inputted in the segmentation model for segmentation processing. Taking porosity defect as an example, it is proved that this method can achieve the accurate segmentation. Comparing with other segmentation method, the precision of defect detection has been improved.

关 键 词：X射线 缺陷检测 支持向量机 图像分割 

分 类 号：TG441.7[金属学及工艺—焊接]