为了充分利用源图像重要特征,提出了一种基于迭代导向滤波与多视觉权重信息的红外与可见光图像融合算法.首先,通过一种迭代导向滤波器将输入图像分解为基础层与细节层; 其次,利用边角信息、清晰度与对比度来综合确定二进制权重系数,再选择导向滤波对其优化,进一步去除噪声并抑制伪影的产生; 最后,应用重构准则对基础层与细节层进行组合,得到融合图像.实验结果表明,与其它多尺度分解相比,该方法具有尺度感知特性,可以更好地分离空间重叠的特征,不仅可以使夜视融合图像的细节信息更突出,还能够有效地抑制伪影.

In order to make full use of the important features of source images, an infrared and visible image fusion algorithm based on iterative guided filtering and multi-visual weight information is proposed. Firstly,input images are decomposed into base and detail layers by iterative guided filtering. Secondly,binary weight coefficients are synthetically determined by edge information,sharpness and contrast,which are then optimized by guided filtering to reduce noise and suppress artifacts. Finally,the fused image is reconstructed by the base layers and the detail layers on the basis of restructuring rules. Experiments show that compared with conventional multi-scale decomposition methods, the proposed method can better achieve separation of spatially-overlapped features, which can not only make the detail information more prominent,but also suppress the artifacts effectively.