基于厚壁工件X射线实时成像的焊缝缺陷自动检测

Automatic weld defect detection based on X-ray images of thick-wall workpieces

基于X射线成像的焊缝缺陷自动检测技术对提高工业射线检测的自动化水平具有重要意义。焊缝缺陷在线连续检测的实时性要求较高,随着工件厚度的增加,其焊缝X射线实时图像的信噪比变得很低,使得已有的处理算法难以在满足实时性的同时,有效处理缺陷误检与漏检之间的矛盾。针对这些问题,在分析了传统方法在厚壁工件X射线图像焊缝缺陷自动检测中存在的问题基础上,对传统方法进行了改进,提出了双阈值背景消除法和平行焊接方向波形分析法,然后利用所提出算法之间的冗余性和互补性,融合多种分割结果以解决缺陷误检与漏检之间的矛盾。试验结果表明:所提出的缺陷自动检测方法能够在满足实时性要求的同时,实现缺陷检出,有效避免误检。

The technology of automatic weld defect detection based on X ray imaging plays an important role in improving the automatic level of industrial radiography inspection. The signal to noise ratio of the X ray real time weldment image decreases with increasing weldment thickness, which makes the current method fail to deal with the conflict of reducing false alarms and avoiding missed detections of weld defects while meeting the requirement of on line continuous detection efficiency. Based on the analysis of drawbacks in traditional background subtraction and grey level profile analysis method, information fusion of multiple image segmentation algorithms was developed to detect weld defects. Double threshold background subtraction and grey level analysis parallel to weld direction were proposed with the segmentation results using different algorithms then fused to deal with the conflict of false alarms and missed detections. Experimental results show that the proposed method can meet the requirement of on-line continuous detection efficiency of weld defects and automatically detect weld defects of thick-wall weldments.