Owing to the complex highlighted background interference in medium-wave images and low target signal-to-clutter ratio （SCR） in long-wave images，the performance of visual tasks in a single band is limited.In order to solve such problems，along with the consideration of the inherent difference and complementary information of the target and background in different band infrared images，a fusion algorithm of infrared dual-band images based on the visual state space model （VMamba） is proposed.First，the global perception ability of the VMamba module is used to establish the long-distance dependence between the target and the background，and the feature information of different band images is extracted and effectively integrated.Second，the semantic information of the target and background is combined to improve the algorithm's ability to perceive the salient features of the target.Finally，under the guidance of improving the performance of downstream tasks，a dual-band image fusion evaluation metric，i.e.，background residual （BR），is proposed to measure the suppression effect of the fused image on the background clutter.The test results in the infrared dual-band measured image data set constructed in this paper show that the proposed infrared dual-band fusion algorithm has the abilities of enhancing the target characteristics and suppressing the background clutter.An SCR of 0.636 and a BR of 2.846×10^-3 are obtained.The SCR is 1.605% higher than that obtained from the optimal general algorithm curvelet transform （CVT），and the BR is 1.625% higher than that obtained from the optimal general algorithm，i.e.，multiscale singular value decomposition （MSVD）.