Thermography Sequence Processing and Defect Edge Identification of TBC Structure Debonding Defects Detection Using Long-Pulsed Infrared Wave Non-Destructive Testing Technology
Chiwu Bu,Zhihui Sun,Qingju Tang,Yuanlin Liu,Chen Mei
DOI: https://doi.org/10.1134/S1061830919010030
2019-10-07
Russian Journal of Nondestructive Testing
Abstract:General infrared image processing and analysis methods mainly include non uniformity correction, enhancement, noise reduction, segmentation, etc. The handling objects of these algorithms are ordinarily single images, which can improve the display effect of defects in infrared image and increase the signal-to-noise ratio. The active heat source excitation is usually used for defects detection in long pulse infrared thermal wave testing technology, so the collected image sequences are dynamic images. At the same time, as in-homogeneous characteristics of thermal excitation, the detection accuracy of defects is affected. Compared with ordinary infrared images, thermal dynamic image sequence is affected by uneven thermal excitation and wave flow. So the characteristics of the dynamic sequence should be analyzed, in order to get better defect display effect. Using the SVD (singular value decomposition, SVD) method for infrared image sequence reconstruction, and the component of surface temperature signal feature information has been extracted. Compared to the original image, the contrast and signal-to-noise ratio of the reconstructed image has been improved. Canny, LOG and other classical edge detection operators were used to detect the edge of infrared images, and their detection results were compared and analyzed. On the basis of analyzing the effect of the classical edge detection operators on the defect edge recognition in infrared images, a hybrid algorithm for edge detection based on Retinex-watershed-Canny operator is proposed for image edge detection, and the thermal barrier coating structure debonding defects' edge has been recognized. A large number of useless and untrue boundary information around has been reduced, and the boundary contour of defects is more clear and continuous, which can improve the feature extraction effect of defects. The research work of this paper lay a certain foundation for the quantitative detection of defects.
materials science, characterization & testing