Abstract: When using an infrared thermograph to detect fire damage to concrete structures, interference comes from a variety of environmental factors, causing an unstable imaging effect, low SNR, and low contrast. Therefore, a single image cannot accurately reflect the internal defects in the concrete. To solve this problem, this paper proposes an image fusion method based on multiresolution singular values. The existing singular value analysis algorithm for multiresolution improvement is recursive, resulting in different resolutions. The components of the multiple infrared thermograph were divided into concrete and a variety of detailed components. For different components, we used different strategies of image fusion as well as the new infrared thermograph. This was to study the annealing algorithm, enhance the image processing, and improve the image clarity and identification rate. Experimental results show that this algorithm shows improvement in definition, spatial resolution, root mean square error, and peak signal-to-noise ratio and provides a reliable basis for the nondestructive testing of fire-damaged concrete.