基于引导滤波慢特征热成像的复合材料缺陷检测  

Defect Detection in Composites Based on Guided Filtered⁃Slow Feature Thermography

作　　者：钟昕 刘凯新 陈平[1,2] ZHONG Xin;LIU Kaixin;CHEN Ping(School of Information and Communication Engineering,North University of China,Taiyuan 030051,China;State Key Lab for Electronic Testing Technology,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学信息与通信工程学院,山西太原030051 [2]中北大学电子测试技术国家重点实验室,山西太原030051

出　　处：《测试技术学报》2025年第2期172-179,共8页Journal of Test and Measurement Technology

基　　金：国家优秀青年科学基金资助项目(62122070);国家自然科学基金青年基金资助项目(62301507);山西省基础研究计划资助项目(202303021222094)。

摘　　要：复合材料在生产制造过程中容易产生缺陷,通过无损检测技术来评估其安全性是必要的。红外热成像技术有着成本低廉和易于设置的优点,然而,红外热成像单次测试采集的多张热图像常含有环境噪声和不均匀加热背景,肉眼直接检视热图像中的缺陷非常困难。鉴于此,提出一种用于复合材料缺陷无损评估的引导滤波慢特征热成像方法。首先采用引导滤波算法对采集的热图像数据进行平滑降噪,然后对滤波后的热图像时间序列进行慢特征分析,以提取其时间上变化最慢的特征(时间序列数据中不变或缓慢变化的特征),从而增强热图像缺陷的可识别性。在碳纤维增强聚合物热图像数据集上的测试结果验证了所提方法的可行性和有效性。Composites are prone to defects during the manufacturing process and need to be assessed for their safety by non-destructive testing(NDT)techniques.Infrared thermography(IRT),with its advan-tages of low cost and easy setup,has become one of the popular techniques for NDT of composite materi-als.However,multiple thermal images acquired by IRT in a single test often contain ambient noise and an inhomogeneous heating background.Therefore,it is difficult to directly visualize defects in thermo-grams with the naked eyes.In this work,a guided filtered-slow feature thermography(GF-SFT)method for nondestructive evaluation of composite defects is proposed.Firstly,a guided filter algorithm is used to smooth thermal images for noise reduction.Then the filtered thermal image time series is subjected to slow feature analysis in order to extract the slowest change in time(the features that are unchanged or slowly changing in the time series data),so as to identify defects.The feasibility and effectiveness of the GF-SFT method are validated by test results on a carbon fiber-reinforced polymer.

关 键 词：无损检测 红外热成像 碳纤维增强聚合物 引导滤波 慢特征分析 

分 类 号：TN219[电子电信—物理电子学] TP391.4[自动化与计算机技术—计算机应用技术] TB33[自动化与计算机技术—计算机科学与技术]