针对含裂纹缺陷的红外热像图对比度差、信噪比低的特点，提出了一种涉及噪声消除、图像细节增强、区域分割等多个方面的红外图像处理算法。算法首先通过时域有限冲激响应（FIR）滤波去除了图像中的部分噪声，然后利用提升小波变换与模糊逻辑非线性算子相结合的方法实现了对红外图像的增强，最后采用改进的区域生长方法对图像中的裂纹区域进行了分割处理。实验结果表明，本文算法有效地滤除了噪声干扰，提高了裂纹图像的对比度，同时很好地保留了整条裂纹中各小段灰度等级不同的特征，为后续测量裂纹深度奠定了必要的基础。

Infrared thermal images with crack defect have the characteristics of low contrast and low signal-to-noise ratio. This article proposes an algorithm based on noise reduction and image detail enhancement and region segmentation to process the infrared images. Firstly, finite impulse response（FIR） filter is adopted to minimize noise of infrared images. Then, the second-generation wavelet transform combined with fuzzy logic nonlinear enhancement operator is used to strengthen image outline. Finally, a modified region growing method is applied to segment crack region from the enhanced image. A group of experiment results demonstrate that the algorithm reduces the noise effectively and improves the contrast of crack image. At the same time, the algorithm accurately reserves the characteristic that each section of the whole crack has different gray values, which can help to measure depth of the crack quantitatively in the future.