目的 鉴于非下采样剪切波变换NSST的红外与可见光图像融合的结果存在细微特征缺失问题，提出一种基于NSST和SWT的红外与可见光图像融合算法，以提升融合图像的质量。方法 首先分别对红外与可见光图像进行NSST分解，各得到一个低频系数和多个不同方向、尺度的高频系数。然后低频系数分别通过SWT分解得到新的低频系数和高频系数，通过SWT分解得到的新的低频系数和高频系数分别采用采用线性加权平均法和区域平均能量取大的融合策略，融合结果再进行SWT逆变换得到低频系数融合结果。高频系数采用区域平均能量取大的融合策略进行融合。最后通过NSST逆变换得到最终的融合图像。结果 通过仿真实验结果表明，文中算法与NSST，SWT和NSCT等算法相比，融合图像在主观视觉上的红外目标更突出，图像细节更清晰，且在IE, AG, QAB/F, SF和SD等评价指标上也最优。结论 文中算法的融合结果能更好地表现源图像的目标信息和细节纹理信息，表明该算法具有优越性。

The work aims to propose a method of infrared and visible image fusion based on NSST and SWT for the missing of fine features in the fusion results of infrared and visible images based on NSST, so as to improve the quality of fused image. Firstly, the infrared and visible images were decomposed by NSST to obtain a low frequency coefficient and multiple high frequency coefficients in different directions and scales, respectively. Then the two low frequency coefficients were decomposed by SWT to obtain the new low frequency coefficients and the high frequency coefficients respectively. The new low frequency coefficient and high frequency coefficient obtained by SWT decomposition separately adopted the linear weighted average method and the region average energy to increase the fusion strategy. The fusion result of low frequency coefficient was obtained by SWT inverse transform, and the high frequency coefficient was fused by the fusion strategy with large region average energy. Finally, the final fusion image was obtained by NSST inverse transform. The simulation fusion results showed that the proposed algorithm was more obvious in visual infrared target and clearer in detail and was also the best in evaluation index such as IE, AG, QAB/F, SF and SD, when compared with NSST, SWT, NSCT, et al. The fusion result of this algorithm can better express the target information and the detailed texture information of the source images, so this algorithm has the superiority.