For the visible and infrared images, dual-band image fusion enhancement is achieved with visual attention extraction under multiple windows. The visual attention map is obtained using local frequency tuned idea, and this attention map simulates human visual system to give different visual weights to different areas and pixels. The method for multiple window-based visual attention maps extraction is designed, which could highlight feature information with different sizes. Based on multi-scale visual attention maps, dual-band image fusion enhancement is achieved. Using dual-band image of demonstration experiments, and combining with both subjective and objective assessment to evaluate various fusion methods. The experimental results demonstrate that the proposed method can efficiently reserving details and meet the need for human visual at the same time, and has faster operating rate.