说明：  针对低可见光图像和红外图像的特点,提出一种基于DT-CWT的自适应图像融合算法.该算法具有好的平移不变性和方向选择性,更适合于人类视觉.先对源图像作双树复小波变换,充分考虑各尺度分解层的系数特征,对 低通子带引入免疫克隆选择,根据统计评价准则定义亲和度函数,自适应获得最优融合权值 对高通子带则根据人类视觉特性定义局部方向对比度,并作为融合准则,突出和增强了各源图像的对比度与细节信息.实验结果表明: 与基于小波的融合结果相比较,本文的融合算法自适应性和鲁棒性更强,较好地保护和显示了源图像中的边缘和细节信息,对比度和清晰度都有所提高.
(Aiming at the characteristics of low visible light images and infrared images, a novel adaptive image fusion scheme based on DT-CWTwas presented. The technique has good shift-invariance and directional selectivity, and ismore suitable for human vision. The DT-CWTwas firstly used to perform amultiresolution decomposition of source images. By taking advantage of the characters of the coefficients, the immune clonal selection algorithm was introduced in low-pass sub- bands and almost optimal fused weightswere obtained adaptively. To high-pass subbands, the local directive contrastwas defined, whichwas based on human visual system. And then the contrastof fused imageswas enhanced and the detail in- formation of source imageswas protected. The experimental results show thatour fused technique is effective and the fusedimages have a better visual quality than theirwavelet counterparts. )