中文摘要：

针对传统基于非下采样Comourlet变换和脉冲耦合神经网络的图像融合方法易出现图像失真的缺点，本文提出一种基于小波变换与PCNN补偿的NSCT域内红外与可见光图像融合方法。首先将红外和可见光图像分别进行NSCT分解，得到低频分量和高频分量；然后对低频分量进行二维小波分解，得到1个低频子带和3个方向子带，对其低频子带采用局部能量加权的方法进行融合，其余3个子带采用绝对值取大的方法进行融合；NSCT分解的高频子带融合规则分为对最高层的融合和其他层的融合，最高层采用绝对值取大的方法进行融合，而其余层采用的是基于改进型的PCNN的方法进行融合；最后将得到的低频子带和高频子带进行NSCT重构获得融合图像。合成及真实图像集实验结果表明，本文算法相对于传统的融合方法增加了图像的纹理和细节信息，有效地抑制了图像失真问题，具有较高的融合精度与较快的融合效率。

英文摘要：

A novel infrared and visible image fusion method was presented based on wavelet transform and PCNN in NSCT domain according to the problem of image distortion which usually caused by the traditional non sampling contoudet transform. Firstly, the low frequency sub-bands and high frequency sub-bands of the infrared and visible image could be obtained by NSCT. The low frequency sub- bands and directional sub-bands of the infrared and visible images could be obtained from the obtained low frequency sub-band by using wavelet transform. Secondly, the proposed weighted local energy method was employed to fuse the acquired low frequency sub-bands, and the maximum method was projected to fuse the acquired directional sub-bands. Thirdly, the maximum method was still employed to fuse the highest level of the high frequency sub-bands obtained by the NSCT, and the improved PCNN was projected to achieve fusion of the other levels of the high frequency sub-bands. Finally, the fusion image could be obtained by integrating the acquired low frequency sub-bands and high frequency sub-bands with the NSCT. Experiments with the synthetic and real image sets showed that the proposed method could better express the texture and detail information of the fusion images compared with the traditional fusion methods, which also could address the problem of the image distortion effectively.