基于局部曲面重构的焊缝X射线图像缺陷分割技术  

Background Subtraction Segmentation of Weld Defects in X-ray Image Based on Partial Surface Reconstruction

作　　者：侯润石[1] 都东[1] 邵家鑫[1] 王力[1] 常保华[1] 

机构地区：[1]清华大学机械工程系先进成形制造教育部重点实验室,北京100084

出　　处：《无损检测》2008年第8期533-535,共3页Nondestructive Testing

基　　金：国家自然科学基金资助项目(No.50628506)

摘　　要：背景消除是焊缝X射线自动检测中缺陷分割的常用方法,其核心是焊缝背景图像的获取。由于焊缝的多样性,常用的低通滤波和曲线拟合方法所得到的背景不够准确,造成较多误检率。研究了基于局部曲面重构的背景消除法,先对焊缝列灰度进行分析,获取潜在的缺陷区域,再对该区域进行曲面重构,获得焊缝背景图像。试验表明,该方法可有效减少误检和漏检,并获得良好的焊缝缺陷形貌。Background subtraction is often adopted in automatic weld defects inspection techniques based on real-time X-ray images. The key to this method is estimating of weld background, and it is mainly realized by low-pass filter or curve fitting methods. Generally, there are many false alarms and inaccurate defect shape due to inaccuracy of weld background obtained by these methods. In this paper a new background subtraction based on partial surface reconstruction was proposed. Firstly, the possible defect positi...

关 键 词：射线检测 实时检测 焊缝缺陷 背景消除法 

分 类 号：TG441.7[金属学及工艺—焊接]