基于区域能量的NSST域偏振图像融合算法

针对传统的偏振图像融合方法存在图像细节丢失、边缘模糊、对比度下降等不足,提出了一种基于区域能量的非下采样剪切波变换域(NSST)偏振图像融合方法。首先,利用NSST对源图像进行分解,获取源图像的低频子带系数和高频子带系数;然后,对分解得到的低频系数基于区域能量加权融合,对高频系数先基于区域能量取大融合,再应用引导滤波进行细节增强;最后,通过NSST逆变换得到融合图像。实验结果表明,融合后的偏振图像能够较好地保留细节信息,同时使边缘更加清晰,图像标准差、平均梯度、信息熵、对比度、空间频率相对传统方法均有不同程度的提高,具有更好的视觉效果。

usion of Polarization Images Based on Area Energy in NSST Domain

In view of the fact that the quality of a fused polarization image inclines to suffer from the halo artifacts such as lost detail information, blurred edge and decreased contrast with a traditional method, this paper has proposed a new polarization image fusion method based on area energy in non subsampled shearlet transform(NSST) domain. Firstly, the source images were decomposed by NSST, and the low frequency subband coefficients and the high frequency subband coefficients of source images were obtained. Secondly, the low frequency coefficients obtained from the decomposition were fused based on the fusion rule of area weighted energy, and the high frequency coefficients were fused based on the fusion rule of area maximum regional energy.Then the guided filter was used for detail enhancement. Finally,the fused polarization image was obtained through inverse NSST. The results of simulation experiment demonstrated that compared with traditional methods the fused polarization image with the method proposed in this paper had clearer edges and better visual effect while preserving the detail information, and the standard deviation, information entropy, average gradient , image contrast and space frequency were largerby different degrees according to objective evaluation.