The existence of blind pixels and non-uniform noise will lead to a significant decrease in the imaging quality of infrared images. Aiming at this problem, a non-uniformity correction algorithm for infrared images is designed. First, the principle of the two-point correction algorithm is introduced. Then the definition of blind pixels and the traditional commonly used blind pixel detection method and blind pixel compensation method are analyzed. On this basis, a gradient threshold blind pixel detection method is proposed, that is, the judgment threshold of the blind pixel is obtained by calculating the grayscale difference between all adjacent pixels. Subsequently, the improved neighborhood substitution method is used for blind pixel compensation, and the above algorithm is used in a self-developed mid-wave infrared camera. Finally, an imaging experiment of a blackbody is designed, and the blind pixel detection and compensation algorithms proposed in this paper are compared with the commonly used methods at this stage, and the imaging quality and non-uniformity indicators of the images before and after correction are compared. The results show that the non-uniformity correction algorithm proposed in this paper can effectively suppress blind pixels and noise, the non-uniformity of the corrected image is reduced by 65%, and the image quality is significantly improved, which can meet the working requirements of the self-developed mid-wave infrared camera.