报告详情
  1. 活动首页
  2. 报告清单
  3. 报告详情
Defect-Aware and Physics-Guided Super-Resolution for Terahertz Nondestructive Evaluation of GFRP Composites
编号：79 访问权限：仅限参会人 更新：2026-09-23 10:45:53 浏览：1次 口头报告

报告开始：暂无开始时间（Asia/Shanghai）

报告时间：暂无持续时间

所在会场：[暂无会议] [暂无会议段]

暂无文件

摘要
Terahertz (THz) imaging is essential for nondestructive testing of GFRP composites, yet diffraction limits cause subsurface defects to appear blurred and low-contrast. Existing super-resolution (SR) methods, optimized for global pixel fidelity, tend to over-smooth backgrounds and either miss sparse defects or generate nonphysical artifacts. We propose a defect-aware, physics-guided SR framework (DAPG-SR) with a dual-head residual U-Net that jointly reconstructs high-resolution images and predicts defect saliency. Two physical constraints are integrated: a degradation consistency loss ensures fidelity to the THz measurement process, and a Laplacian-based PDE loss preserves sharp boundaries. A defect-oriented sampling and weighting strategy further prioritizes critical flaw regions. Experiments on GFRP THz images show that while conventional models achieve marginally higher global metrics, our method delivers superior defect visibility with sharper edges and fewer artifacts. This work offers a reliable, interpretable SR solution for automated composite inspection in AI-driven sensing applications.
 
关键词
Terahertz Imaging,Super-Resolution,Defect Inspection,Physics-Guided Deep Learning,Non-Destructive Testing,GFRP Composites
报告人
Guoyu Zhong
Mr. Shenzhen University

稿件作者
Guoyu Zhong Shenzhen University
Wenlong He Shenzhen University
Mn Zhai Shenzhen University
发表评论
验证码 看不清楚，更换一张
全部评论
重要日期

  • 会议日期

    11月06日

    2026

    至

    11月08日

    2026

  • 10月15日 2026

    初稿截稿日期

主办单位
IEEE Instrumentation and Measurement Society
承办单位
Sichuan University
联系方式
登录查看完整联系方式
历届会议
移动端
在手机上打开
小程序
打开微信小程序
客服
扫码或点此咨询