Aiming at the problems of unclear texture, low brightness and high noise of infrared images, an adaptive multi-feature fusion algorithm for infrared image enhancement is proposed in this paper. Firstly, the automatic linear mapping method is used to extract 16-bit infrared images from 14-bit infrared images, which improves the visual effect. Secondly, the combined processing method of GUM and MSRCR is introduced to obtain effective information on different scales of the image and improve the contrast of the image. Finally, the adaptive weight map is designed and combined with the characteristics of the image pyramid structure to complement and fuse the effective information of different feature layers, which improves the brightness of the image and enriches the texture information of the image. Experimental results show that this algorithm can effectively improve the contrast and visual effect of infrared images, and its AG is increased by about 0.6% compared with the existing algorithms. The PSNR is about 10% higher, the image edge information efficiency is about 11% higher, and the image sharpness is about 10% higher.