中国机械工程 ›› 2010, Vol. 21 ›› Issue (1): 59-62.

• 信息技术 • 上一篇    下一篇

工字形焊件射线图像中微小缺陷的分割及提取

石端虎1;刚铁2;张华军3;杨根喜1
  

  1. 1.徐州工程学院,徐州,221008
    2.哈尔滨工业大学现代焊接生产技术国家重点实验室, 哈尔滨,150001
    3.哈尔滨理工大学, 哈尔滨,150040
  • 出版日期:2010-01-10 发布日期:2010-01-22
  • 基金资助:
    江苏省高校自然科学研究项目(09KJD430011)
    Jiangsu Provincial Natural Science Research Program of Higher Education of China(No. 09KJD430011)

Segmentation and Extraction of Small Defects in X-ray Images for I Style Weldments

Shi Duanhu1;Gang Tie2;Zhang Huajun3;Yang Genxi1
 
  

  1. 1.Xuzhou Institute of Technology, Xuzhou,Jiangsu, 221008
    2.State Key Lab of Advanced Welding Production Technology, Harbin Institute of Technology, Harbin, 150001
    3.Harbin University of Science and Technology, Harbin, 150040
  • Online:2010-01-10 Published:2010-01-22
  • Supported by:
    Jiangsu Provincial Natural Science Research Program of Higher Education of China(No. 09KJD430011)

摘要:

采用X射线实时成像系统对工字形激光焊件进行检测,得到了含有微小缺陷的低对比度射线检测图像。在分析图像灰度分布特征的基础上,采用数学形态学方法进行背景模拟,并结合背景相减、自动阈值算法提取出了检测图像中的微小缺陷。在结构元尺寸的确定方面,提出了线灰度分布曲线拟合并相减,再搜索相减结果最大值的方法。试验结果表明:该方法能很好地实现检测图像中微小缺陷的分割,且缺陷分割保真度高,算法适应性强。该方法只对感兴趣的区域进行处理,显著提高了图像处理的速度。

关键词:

Abstract:

X-ray real time image system was used to detect I style laser weldments nondestructively, and X-ray detection images were obtained. On the basis of image gray feature analysis,a mathematical morphology method was adopted to simulate background, combining the methods of background subtraction and automated threshold, the small defects were extracted from the detected images. The method of curve fitting and searching the maximum value of subtraction results was put forward to determine the size of structural elements. The experimental results show that the extracted method can achieve the segmentation of small defects well, defects fidelity is high, and the adaptability of algorithm is powerful. In addition, the above algorithm is only carried out in the area of interest, so the speed of image processing will be improved greatly.

Key words: I style, laser weldment, X-ray image, small defect, defect segmentation

中图分类号: