针对红外偏振融合后图像边缘模糊不清和细节信息不明显等问题,本文提出了一种基于局部能量匹配的红外偏振图像融合方法。首先,融合偏振角图像和偏振度图像得到偏振特征图像;然后运用非下采样剪切波变换(NSST)把偏振特征图像和光强图像分解为低频子带和高频子带,运用顶帽(Top-hat)变换处理偏振图像的低频信息,提取目标;最后采用基于局部能量匹配和局部方差相结合的融合规则融合低频子带和高频子带。实验结果表明,本文算法与 NSCT算法相比较,融合后图像的整体互信息值提高了 8.7%,方差提高了 3.9%,很好的保留了图像的细节信息。

In order to highlight the edge of infrared image and the detail information, the article proposes a fusion method based on local energy matching for IR polarization and light intensity images. First, The polarization characteristic image is the result of the fusion of AOP and DOP images, then polarization characteristic image and infrared image are decomposed into low frequency and high frequency sub-bands by using the NSST, and the low frequency information of polarization image is processed by using Top-hat transform, with target extracted. Finally the low-frequency sub-band and high frequency sub-band are fused by different fusion rules which are based on local energy matching and local variance. Experimental results show that compared with the NSCT algorithm, the mutual information value is increased by 8.7% and the variance is increased by 3.9% and the image details are retained very well.