Fusion of near-infrared and visible images based on saliency-map-guided multi-scale transformation decomposition

In this research, we propose a near-infrared (NIR) and visible (VIS) image fusion method based on saliency-map-guided multi-scale transform decomposition (SMG-MST) to solve the problem of color distortion. Although the existing NIR and VIS image fusion methods can enhance the texture information of the fused image, they cannot control the scattering of light from objects in the fused image resulting in color distortion. The color distortion region usually has good saliency, so using saliency map to solve the above problem is a good choice. In this paper, a visible image guided by saliency map is introduced in the low frequency part, which can weaken the scattering of too much light from objects in the image. In addition, the local entropy of the NIR is used to guide the visible photon images, so the results contain more details. Both qualitative and quantitative experiments demonstrate the effectiveness of our algorithm, and the comparison of algorithm running times shows the high efficiency of our method.