中文摘要：

针对传统的融合方法对于可见光图像丰富的背景信息不能有效地保留到融合图像的问题，本文结合像素级和特征级融合的优势，提出了一种新的图像融合算法．首先，对来自同一场景配准后的红外与可见光图像进行NSCT变换；其次，在保留红外图像中目标信息的基础上，出于尽可能比较完善地将可见光的背景信息融入到融合图像的目的，对于低频图像采用基于区域能量比的自适应加权平均的方法；再次，考虑到低频图像和高频方向子带中包含大量的边缘细节信息，本文在Petrovic多层级融合方法的基础上进行相应的改进，利用低频子带边缘特征的融合以指导高频方向子带系数的融合；最后，通过NSCT逆变换得到融合图像．对多组红外与可见光图像进行实验，并对融合图像进行主客观评价，实验结果表明，该算法得到的融合图像具有非常好的主观视觉效果和客观量化指标，融合性能显著地优于传统的融合方法．

英文摘要：

Specific to the drawback that classical methods are not effectively for retaining the abundant background information of visible image into fusion image, combines with the superiorities of signal level and feature level fusion, a novel fusion algorithm is proposed. Firstly, the registered infrared and visible images from the same scene were transformed by NSCT transforms. Secondly, on the basis of reserving the target information of infrared image, for the purpose of integrating as much more visible image infor- mation into fusion image as possible, the low frequency coefficient is fused by adaptive weighted average based on regional energy ratio. In additon, taking into account that low and high frequency coefficient are plentiful of edge details, the paper enrichs Petrovic fusion system with more edge components, and then the high frequency coefficient fusion is conducted by using the edge feature fusion of low frequency coefficient. Finally, the target image is obtained by performing inverse NSCT transforms. The algo- rithm has been used to merge multiple sets of infrared and visible images, experimental results indicatethat the fused image obtained by the proposed method has a much better subjective visual effect and ob- jective evaluation criteria, fusion performance is dramatically better than traditional fusion methods.