利用NSCT域邻域特性区域化方法融合红外与微光图像

针对同一场景的红外与微光(可见光)图像融合问题,提出了一种利用邻域特性区域化处理的非下采样Contourlet变换(NSCT)融合方法.首先,对红外和微光源图像进行多尺度、多方向分解;然后对低频系数采用邻域能量上改进的区域化加权融合规则,高频系数采用邻域能量区域化匹配的系数选择方案与邻域方差上的区域方差取大融合规则;最后利用NSCT反变换进行重构得到融合图像.实验结果表明,本文方法信息熵略低于亮度过度增大的参考文献中的方法,由于传统方法引入虚假细节信息导致空间频率略高于所提方法,但其互信息与边缘保持度(Q)指标值较好,其融合图像整体效果优于对比方法.本文方法在保留源图像主要信息及捕捉细节信息方面效果显著,且融合的图像具有较好的视觉效果.

Fusing Infrared and Low Light Level Images by Using the Method of Neighborhood Characteristics Regionalization in Domain

A fusion algorithm for infrared and low-ligh level(visible) images based on neighborhood characteristic and regionalization in NSCT domain was proposed. Firstly, the NSCT was performed on the infrared and visible images at different scales and directions. The low-frequency coefficients were fused with a rule of an improved regional weighted fusion method based on neighborhood energy, and the high-frequency coefficients were fused with a rule of an improved neighborhood energy and regionalization coefficients options and an area variance chooses max based on a neighborhood variance regionalization rule. Finally, the fused coefficients were reconstructed to obtain the fused image. Experimental results show that the information entropy of proposed method is lower than that of the method in the reference whose luminance increases excessively. The introducing of false details of traditional methods results in spatial frequency is slightly higher compared with the proposed method. But the mutual information and edge retention(Q)index value of the proposed method are better and the fusion image is superior to other contrast methods. The proposed method presented better effects in retaining the source images information and capturing details, and fused image had better visual effects.