脉冲红外热波检测的图像序列处理与缺陷边缘识别  

Image sequence processing and defect edge identification in pulsed infrared thermal wave detection

作　　者：唐庆菊[1] 刘元林[1] 梅晨[1] Tang Qingju Liu Yuanlin Mei Chen(School of Mechanical Engineering, Heilongjiang University of Science ＆ Technology, Harbin 150022, China)

机构地区：[1]黑龙江科技大学机械工程学院,哈尔滨150022

出　　处：《黑龙江科技大学学报》2017年第2期200-204,209,共6页Journal of Heilongjiang University of Science And Technology

基　　金：国家自然科学基金项目(51403139);黑龙江省省院科技合作项目(YS15A10)

摘　　要：针对脉冲红外热波检测热障涂层结构脱粘缺陷构件时,所采集红外图像噪声较大和边缘模糊的问题,采用SVD奇异值分解算法对红外图像序列进行重构,实现了构件表面温度信号特征信息的提取。与原始热图相比,处理后的图像提高了对比度和信噪比。采用Canny、LOG等经典检测算子对缺陷进行边缘识别,并在分析其检测效果的基础上,提出了基于Retinex-分水岭-Canny算子的边缘检测混合算法,实现了热障涂层结构脱粘缺陷边缘的有效识别。This paper is aimed at overcoming a greater noise and fuzzier edge in the infrared images, as occurs in the detection of thermal barrier coating debonding defects. The research involves reconstructing the infrared image sequence using singular value decomposition algorithm; extracting the feature information of component surface temperature signal ; improving the contrast and signal-to-noise ratio of the processed image relative to the original image ; detecting the edge of defects using Canny, LOG and other classical detection operators; and based on the analysis of the detection results, proposing a hybrid edge detection algorithm based on Retinex-watershed-Canny operator and thereby achieving an effective identification of the edge of thermal barrier coating debonding defects.

关 键 词：红外图像 奇异值分解 边缘检测 混合算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]