红外与可见光图像快速融合算法  

A Fast Fusion Algorithm of Visible and Infrared Images

作　　者：钱小燕[1] 韩磊[2] 王帮峰[1] 

机构地区：[1]南京航空航天大学民航学院,南京210016 [2]南京华为技术有限公司,南京210000

出　　处：《计算机辅助设计与图形学学报》2011年第7期1211-1216,共6页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(10772081)

摘　　要：为了增强融合效果,提高融合速度,提出一种快速有效的红外与可见光图像融合算法.首先采用局部直方图均衡和中值滤波对红外与可见光图像进行对比度增强和去噪处理;然后对增强后图像进行线性融合,生成伪彩色融合图像;最后简化Lab颜色空间与RGB颜色空间之间的转换模型,并在简化的Lab空间实现全局颜色迁移,将可见光参考图像的一阶统计信息融合进伪彩色图像,获得自然感的真彩色图像.对比实验结果表明,采用文中算法生成的融合图像更加清晰、自然,速度更快、更简单.To improve both the quality and performance speed,a simple and fast fusion approach for night color vision images is presented in this paper.Firstly,the contrast of infrared and visible images to be fused is adjusted with local histogram equalization.Then the two enhanced images are fused into three components of an RGB image with a simple linear fusion strategy.To obtain false color images possessing a natural day-time color appearance,this paper proposes an approach which transferred color from the reference to the fused images in a simplified Lab space.In this way,inappropriate color mappings can be avoided and overall discrimination capabilities are enhanced.Experimental results based on two different data sets show that the final scene has a natural day-time color appearance and the fusion process is simple and fast.

关 键 词：图像融合 颜色传输 局部直方图均衡化 红外图像 颜色空间 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]