The image details and contour information cannot be fully reflected for the current infrared single-band data. It is difficult for the weak-small target to resist background interference after imaging, so that the image produces a low ratio of signal-to-noise. Therefore, it is necessary to use the texture difference of different band data to improve the signal-to-noise ratio of the image by using the complementary fusion method. Based on the above-mentioned, a fusion method based on wavelet transform and feature extraction is proposed. Firstly, the source images are multi-scale and two-dimensionally decomposed to obtain low-frequency information and high-frequency information. And that, the high-frequency information adopt the method of maximizing the absolute value, the low-frequency information adopt the method of weighted averaging, and reconstruct the image. Then, the infrared feature extraction method is used to obtain the medium wave and long wave feature images. Finally, the reconstructed image is contrast-modulated and refused with the medium-long wave infrared feature image. The fusion results are compared with a variety of fusion algorithms. The experimental results show that the algorithm can enhance the gray scale of weak-small targets in the image, which can identify the target well and solve the problem of weak target against background interference in infrared images.