一种红外偏振与光强图像的暗原色多特征分离融合方法

摘要

针对现有红外偏振和光强图像融合算法不能同时兼顾提高对比度、突出明亮特征和保持边缘细节信息的问题,提出一种暗原色多特征分离融合的新方法.首先,对源图像进行局部最小值滤波和引导滤波得到暗原色图;然后,经过差值、绝对值取小融合的处理,提出暗原色多特征分离方法,得到包含亮、暗以及边缘细节信息的三部分图像;最后,对三部分图像采用不同规则分别融合,通过重构和亮度调整得到最终融合图像.实验结果表明,本文方法能有效地融合源图像的互补信息,不仅能够保留边缘细节信息,同时也能突出明亮特征和图像整体对比度,在实际的目标识别中具有一定的优势.%In order to consider enhancing contrast of images,highlighting bright feature and keeping edge detail information in current infrared polarization and infrared intensity images fusion algorithm simultaneously,a new method of multi-feature separation fusion of dark channel map is proposed.Firstly,the dark channel maps are obtained by the local minimum and guide filtering form the source images.Then,by the subtraction and taking the minimumfusion,a new method of multi-feature separation of dark channel maps are proposed to get the three part images that contain light,dark,and edge detail information.Finally,the three part images are fused through different rules,andthe final fusion image is obtained by reconstruction of the image and brightness adjustment.Experiments show that the proposed method can effectively merge the complementary information form the source images,which not only KEEP the edge detail,but also highlights bright feature and contrast of the image,and has certain advantages in the actual target recognition.