基于BEMD和自适应PCNN的偏振图像融合

Polarization Image Fusion Based on BEMD and Adaptive PCNN

针对传统偏振图像融合视觉效果较差,纹理细节信息保留不充分等问题,结合BEMD变换的多分辨率、多尺度特性和PCNN全局耦合、脉冲同步激发等优点,提出了一种基于BEMD和自适应PCNN的偏振图像融合算法。首先融合线偏振度图像和偏振角图像得到偏振特征图像,然后对偏振特征图像和强度图像进行BEMD分解,对于低频分量,采用局部能量的融合方法;对于高频分量,采用区域方差自适应调整PCNN的链接强度,根据其显著性度量融合各子带系数。实验结果表明,该算法在视觉效果和空间频率,边缘保持度,互信息和平均梯度等多项评价指标上较其他算法均有优势。

In order to solve the problems of traditional polarization image fusion such as poor visual effect and in- sufficient texture detail information, this paper combines the advantages of muhi-resolution, multi-scale characteristieof BEMD transform and global coupled and pulse synchronization excitation of PCNN, a new algorithm for polarization im- age fusion of BEMD and adaptive PCNN was proposed. The polarization characteristic image is obtained by integrating the linear polarization image and the polarization angle image. Then, the polarized characteristic image and the intensity image are decomposed by using BEMD. For low frequency components, the local energy fusion method is adopted. For high frequency components,the local variance is used to adjust the link strength of PCNN, and the sub-band coeffi- cients are fused according to the significance measure. The experimental results show that the proposed algorithm has advantages over other algorithms in terms of visual effects, spatial frequency, edge preserve, mutual information and av- erage gradient.