基于伪维格纳分布的图像融合

提出了一种基于伪维格纳分布(PWVD)的融合方法.利用一维N像素的滑动窗口在各个方向上对各待融合图像进行伪维格纳变换,选择均方根最大的方向为各待融合网像的PWVD分解方向,分解形成待融合图像不同频段的能量谱图,然后,针对各待融合不同频段的能量谱图,融合原则足高频段取区域能量最大,低频段取能量方差最大,形成具有不同频段的融合能量谱网,最后,对能量谱网进行PWVD逆变换,形成融合图像.对红外与可见光图像、多聚焦图像、电子计算机X射线断层扫描(CT)图像与磁共振(MR)图像和红外与合成孔径雷达(SAR)图像进行了融合实验,并对融合图像和待融合图像进行了信息熵对比.实验结果表明,采用本文算法的融合图像保留了待融合图像的绝大部分信息.

An Image Fusion Algorithm Using Pseudo Wigner Ville Distribution

An image fusion algorithm using pseudo-Wigner Ville distribution (PWVD) is proposed. Firstly, a 1-D pseudo Wigner Ville distribution fuction is applied to nonoverlapping N-pixel window slices of the original image in different orientations, the resultant data is arranged as N energy maps according to the orientation where the mean square root is max. Then, the fusion principle of multiscale decomposition of the source images is the max partial energy of energy map in high frequency and the max variance of energy maps in low frequency. So, the resultant data can be arranged as N energy maps in different frequencies. Finally, the fused image is formed through contrary PWVD to the fused energy maps. The fusion experiment for infrared and visible images, multifocus images, computer X-ray tomography and magnetic resonance images, infrared and synthetic aperture radar images are taken, and the entropy comparison between fused image and source images is given. The results show that the fused image obtained by the presented algorithm preserves the abundant information of the source image.