基于加权最小二乘滤波和引导滤波的铸件DR图像融合
DOI:
CSTR:
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

TH744

基金项目:

国家自然科学基金(61771003)项目资助


Casting DR image fusion based on weighted least squares filter and guided filter
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    为了让两张不同灰度窗宽和窗位的铸件X射线DR图像包含的缺陷呈现在一张图像中,发展了一种基于加权最小二乘滤波和引导滤波(guided filter)的铸件X射线DR图像融合方法。选择其中一张图像作为底层图像,另一张作为细节图像。使用两种具有边缘保护性质的滤波提取细节图中包含的缺陷信息。对底层图像作加权最小二乘滤波作为框架,再用底层图像和框架做差,分离出底层图像本身包含的缺陷信息。利用引导滤波把这两类缺陷信息加权求和。最后用框架承接加权求和后的结果后得到融合后的图像。实验结果表明获得的融合结果能够显示来自不同灰度窗宽和窗位的两张DR图像的共5个缺陷或3张铸件DR图像的共7个缺陷。本文方法可应用于铸件射线无损检测。

    Abstract:

    In order to present the defeels contained in two casting X ray DR images with diflerent grayscale window widths and windowlevels in one image, develops a kind of casting X-ray DR image fusion method based on weighted least squares filter and guided filter,One of the images is selected as the base image and the other image is selected as the detail image. The aforementioned two filters withedge-preserving property are utilized to extract the defeet information contained in the detail image. The weighted least squares filter isapplied to the base image and the output is taken as the frame, then the defects contained in the base image are separated by subtractingthe frame from the base image. The guiding fiter is used to sum the two types of defeet information with weights. Finally, the frame isused to carry the results of the weighted summation to get the fused image. Experiment results show that the obtained fused result candisplay 5 defects of two casting DR images with different grayscale window widths and window levels or 7 defeets of three casting DRimages with different grayscale window widths and window levels. The method presented in this paper can be applied to the radiographicnondestruetive testing of castings.

    参考文献
    相似文献
    引证文献
引用本文

羊肇俊,曾 理.基于加权最小二乘滤波和引导滤波的铸件DR图像融合[J].仪器仪表学报,2021,(6):211-220

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2023-06-28
  • 出版日期:
文章二维码