Infrared and visible image fusion via intensity transfer and direct matrix mapping

Infrared (IR) and visible (VIS) image fusion has turned into a research hotspot in image processing due to various applications recently. In this paper, so as to make the resulted images not only retain the thermal object information in the IR images and the background information in the VIS images simultaneously, but also be more in line with human vision, we propose a novel intensity transfer and direct matrix mapping-based fusion method, which we named IT-DMM. Specifically, we formulate such fusion as a minimization problem, and our solution to the problem is a direct matrix mapping from the input images to the resulted image. The weights used in our IT-DMM, which are computed by using spatial saliency map of the IR image, are vital to the performance of the proposed method and are analyzed in detail. Our IT-DMM has high computational efficiency due to the direct matrix mapping and the efficient weight computation method, and thus it has potential practical value. Qualitative and quantitative experiments on eighteen IR and VIS image pairs indicate that our IT-DMM outperforms eleven state-of-the-art image fusion methods in obtaining perceptually better fusion results with almost the same thermal object information in IR images and background information in VIS images.