脉冲涡流热成像缺陷检测技术可以对导电试件进行快速准确的检测, 但是容易受表面加热不均的影响。采用因子分析法计算了红外图像序列的公因子图像, 对 45#钢的上表面和下表面裂纹进行检测, 并与主成分分析法比较。发现因子分析法能够抑制表面不均匀加热的影响, 扩大裂纹的检测范围, 因子分析法重建的图像质量要优于主成分分析法, 并能结合实际给出合理的解释。在提高计算效率方面, 公因子图像在热像仪采样频率低至 50 Hz时仍然可以有效识别出裂纹, 选取合适的图像序列和公因子数可以减小数据处理量并提高图像质量。

The defects of conductive materials can be detected efficiently and accurately with pulsed eddy current thermography. But the result is interfered significantly by inhomogeneous heating. The factor analysis is applied to calculate the common factor images to detect the surface and subsurface cracks of 45# steel, and then compare with the images reconstructed by principal component analysis. It shows that the inhomogeneous heating can be decreased and the detection range is broadened with the common factor images. Compared with principal component analysis, the quality of the images reconstructed by factor analysis is better, and the factor images can be explained with the corresponding situation. In terms of computational efficiency, the cracks is still detectable by common factor images when the thermal image sampling rate is down to 50 Hz. The selection of appropriate image sequences and the number of common factor is helpful to enhance the quality of reconstructed images.