基于NSST变换域KFE与DUM⁃PCNN的伪彩色图像融合

摘要

针对夜视图像融合过程中,目标信息不突出、色彩信息显示不自然等问题,提出了一种新的基于NSST变换域的伪彩色融合方法,采用Kirsch特征能量(KFE)和所构造的双通道单链接记忆性脉冲耦合神经网络(DUM-PCNN)设计了新融合方案,在NSST变换域内完成红外(IR)与可见光(VIS)的融合;将融合后的图像与IR及VIS进行线性组合并将其分配至YUV彩色空间,合成伪彩色(FC)图像.最后,利用色彩空间的信息转换技术获得染色后的夜视图像.基于不同数据集的对比实验结果表明,文中所提出的FC图像融合方法,能得到突出的目标信息,呈现出与自然图像相近的外观.%The target information is not prominent in the process of night vision image fusion, color information dis?play unnatural. A novle false color image fusion method based on the nonsubsampled shearlet transform ( NSST) is proposed.Using Kirsh feature energy( KFE) and constructing the dual channel unit link memristive pulse coupled neural network ( DUM?PCNN) design the new fusion scheme, completing the infrared ( IR) and visible ( VIS) im?age fusion in the NSST domain. The fused image,IR and VIS carried on the linea combination and it assiged the YUV color space, synthesis the false color image ( FCI) . Finally, Applied color space conversion technology to ob?tain the pseudo?color image. Experimental results based on different data sets show that the proposed the fusion method can obtain the prominent target information and present the similar to natural images color appearance.