Multi-directional shearography for high-precision localization of near-surface defects

Shearography is an effective technique for detecting and localizing near-surface defects in engineering materials. However, as critical experimental parameters in shearography, both the loading magnitude and shearing magnitude can lead to localization errors or even misjudgment. Although existing shearography-based approaches have made progress in defect localization, challenges remain in achieving both high localization accuracy and robustness. To address these challenges, we propose a novel framework combining two key innovations: (1) a multi-directional shearography system to separate and eliminate errors caused by shearing magnitude, and (2) a criterion for optimal loading magnitude selection to suppress errors caused by loading magnitude. Using our method, we performed defect localization on a test object containing three types of defects. Experimental results demonstrate that, within a suitable range of loading magnitude, our method achieves a relative error of 3.6 % in the defect area (indicating size accuracy) and an average intersection over union of 0.8156 (reflecting overlap consistency with ground truth). Furthermore, key parameters of multi-directional shearography are analysis, and defects with extreme aspect ratios are localized, demonstrating the superior performance of our method.