利用搭建的电涡流脉冲热成像检测平台，对常用金属的表面缺陷进行了检测。为了去除热成像图中的噪声，提高金属试件表面缺陷的对比度，采用独立成分分析法分析了表面缺陷的热响应特征。结果表明，独立成分分析法可有效地识别出缺陷、线圈、缺陷及线圈附近、试件边缘等区域的热响应信号特征，缺陷区域成像清晰，实现了金属构件表面线性缺陷和微小开孔的可视化检测。

The built-up eddy current pulse thermal imaging inspection platform was used to detect the surface defects of common metals. In order to remove the noise interference in the thermal image and improve the contrast of the surface defects of the metal specimen, the independent component analysis method was used to analyze the thermal response characteristics of the surface defects. The results show that the independent component analysis method can effectively identify the thermal response signal characteristics of the defect area, the coil area, the defect and the area near the coil, the edge area of the test piece, etc. The visual inspection of openings defect area is imaged clearly, and the linear defect and the small surface of the metal component are realized.