Infrared and Visible Image Fusion via Total Variation and Alternating Direction Method of Multipliers

Owning to the different imaging principles of the infrared (IR) and visible images, IR images contain object information by thermal radiation and visible images contain abundant details of the scene. Image fusion can combine the complementary information of IR and visible image. To obtain ideal fusion result, we should consider the difference between multi-sensor images and use different representations for the source images. In this paper, a new total variation (TV) based fusion method for IR and visible image is proposed. Our method consists of two steps. First, we represent the fusion task as a L 2 −L 1 TV model. The fidelity term is formulated by L 2 norm to constrain the fusion result preserving intensities in IR and visible images, and the regularization term is formulated by L 1 norm to constrain the fusion result preserving gradients in visible image. Second, we adopt the framework alternating direction method of multipliers (ADMM) to solve the TV minimization problem. We take both qualitative and quantitative tests on our method, and make comparisons with eight state-of-the-art methods. The experimental results indicate that our method is superior to the other eight methods.