中文摘要：

脉冲热像检测中直接获取的原始热像往往信噪比较低、对比度较差.为了抑制各种噪声的不良影响,提高脉冲热像检测的缺陷探测能力,运用基于对数多项式回归的图像重建方法对碳纤维增强塑料层压板脉冲热像检测的热像序列进行图像重建与增强.通过对一维热传导模型的解析分析论证了对数多项式回归法的理论基础;通过层压板检测实验对图像重建与增强方法的实际处理效果进行了验证;采用基于统计学的图像评估标准对图像增强效果作了定量评价.结果表明利用三次对数多项式回归即可精确地重建本次实验的原始热像,同时能有效地克服复合材料层压板脉冲热像检测中随机噪声和加热不均效应对缺陷识别的干扰;经过数据重建后所作的数字图像比原始热像有更高的缺陷显示度或分辨力,其中以三次对数多项式回归公式中的二次项系数所作的数字图像的信噪比最大.

英文摘要：

Raw thermal images acquired by pulsed thermography （PT） usually have low signal noise ratio （SNR） and temperature contrast. In order to restrain the bad effects of noise and improve the defect detecting ability of PT, image reconstruction methods based on logarithm polynomial regression were carried out here to reconstruct and enhance PT image sequences of carbon fibre reinforced plastic （CFRP） laminates. The basic theory of these methods were analyzed with the analytic solution of one-dimensional heat transfer model; the actual effects of image reconstruction and enhancement were verified through the laminate testing experiment; the image enhancement effects were evaluated quantitatively by the image evaluation standards based on statistics. The results indicates that the cubic logarithm polynomials can accurately reconstruct the raw thermal images in the experiment, and can effectively suppress the interference to defect recognition caused by random noise and uneven heating in the PT testing of composite laminates; the digital images from data reconstruction have clearer defect display or higher defect resolution than the raw thermal images, and the quadratic coefficient image of cubic logarithm regression shows the highest SNR.