中文摘要：

为在红外和可见光图像的融合过程中减少细节信息的损失,提出一种基于目标提取和信息逼真度的图像融合方法。借助非降采样剪切波变换（non-subsampled Shearlet transform,NSST）,对红外与可见光图像采用多尺度多方向分解策略,得到高低频子带系数;对低频子带采用基于目标特征信息的权值比例分配的融合方法,对高频子带采用信息逼真度和视觉敏感度系数相结合的融合方法;采用NSST逆变换重构获得融合后的图像。对比实验结果表明,使用该算法得到的融合图像可以最大程度保存细节信息,使融合图像具有更好的视觉感。

英文摘要：

To reduce the loss of detail information for the infrared and visible image fusion, an image fusion based on target ex- traction and information fidelity was proposed, and the high and low frequency subband coefficients were obtained using non-sub- sampled Shearlet transform to decompose the infrared and visible image in multi-scale and multi-direction strategy. The fusion method of the weighted average of target feature information was adopted for the low frequency subband components. The fusion method of information fidelity and visual sensitivity coefficient was adopted for the high-frequency components. The fused image was obtained by NSST inverse transform reconstruction. The comparative experimental results show that the improved fusion method can keep the detail information to the greatest degree, making the fused image has better visual sense.