Aiming at the problems of low contrast and unclear scene details in infrared and visible image fusion, an infrared and visible image fusion method based on total variation is proposed. In this method, the difference between the fusion image after each iteration and the visible and infrared images is used as the regular and fidelity terms of the total variational model. At the same time, a non-incremental diffusion function about infrared images is introduced to guide the diffusion and suppress the smoothness of the edge information of the infrared images, so that the image fusion problem can be transformed into an image denoising problem, and finally it is transformed into a functional problem of finding extreme values. The experimental results verify the effectiveness of the method from visual quality and objective evaluation.