红外无损检测图像处理算法的研究

Research on Image Processing Algorithms for Infrared Non-destructive Testing

红外无损检测的原始热图像通常存在信噪比低、温度对比度差的问题。为了排除噪声的不良影响,增强缺陷的对比度,需要对原始热图像进行处理和重构。而算法的选择决定了图像处理的效果。本文通过机械加工模拟腐蚀件的实验,对红外热图像处理中的四种典型算法进行了研究。对处理结果做出了比较,得出了各种算法的适用性和性能对比结论。该结论可作为实际应用中算法选择的依据和参考。

Generally, the original images obtained in infrared Non-destructive Testing （NDT） have the disadvantages of low Signal-to-Noise Ratio （SNR） and poor temperature contrast. To eliminate the harmful effect of noise and increase the contrast of defects, the original images obtained in infrared Non-destructive Testing （NDT） should be processed and reconstructed. However, the effectiveness of the image processing is determined by the selection of algorithms. Four typical algorithms used in infrared image processing are studied in the simulation experiment of machined corrosion parts. The processed results are compared with each other and the applicability and performance conclusion of each algorithm are obtained. The conclusion can provide the guidance for the selection of algorithms in practical infrared image processing.