红外与微光融合图像的多尺度色彩传递算法  

Multi-resolution Color Transfer Algorithm for Fusion Image of Infrared and Low-level light

作　　者：谯涵丹 富容国[1] 王贵圆 

机构地区：[1]南京理工大学近程高速目标探测技术国防重点学科实验室,江苏南京210094

出　　处：《红外技术》2016年第2期157-162,共6页Infrared Technology

基　　金：江苏省产学研联合创新资金(BY2012024);中国兵器工业集团预研项目

摘　　要：针对红外与微光的灰度融合图像不利于人眼观察的问题,提出了一种在YCbCr颜色空间的基于多尺度多分辨率分析的色彩传递算法,帮助观察者获取丰富的场景信息和舒适的观察效果。该算法首先在Y通道将红外和微光图像进行灰度融合;然后将灰度融合图像和彩色参考图像的Y分量做拉普拉斯金字塔分解,比较其金字塔各层系数差异,综合各层系数差异找出最佳匹配点;最后将参考图像中最佳匹配点的Cb和Cr值传输到灰度融合图像中,生成最终的具有自然色彩的融合图像。实验结果表明,本文方法生成的彩色融合图像接近自然真实的颜色,有利于人眼对目标的观察与场景的理解。In order to convert the fusion image of infrared and low-level light images into color image which accords with human visual perception, and helps observers to get rich scene information and comfortable visual effect, a color transfer algorithm based on multi-scale and multi-distinguishability in CbCr color space was proposed. Firstly, infrared and low-level light images were fused in Y channel. Then, the laplace pyramid decomposition was done with the gray fusion image and the reference image. The best matching point was located by comparing the coefficients of each pyramid layer of both images and finding the minimum difference value point. Finally, the Cb and Cr components of the best matching point in reference image were transferred into gray fusion image to get the final fusion image with natural color effects. The experimental result shows that the color-transferred image has a natural color perception to human eyes, which benefits for the human eye observation of the target and understanding of the scene.

关 键 词：色彩传递 图像融合 红外图像 微光图像 YCBCR颜色空间 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]