基于关键帧提取的红外与可见光序列图像快速融合  

The Fast Fusion for Infrared and Visible Sequences Image Based on Key Frames Extraction

作　　者：刘思汝[1] 杨风暴[2] 陈磊[2] 

机构地区：[1]中北大学仪器与电子学院,山西太原030051 [2]中北大学信息与通信工程学院,山西太原030051

出　　处：《红外技术》2014年第5期350-354,共5页Infrared Technology

基　　金：国家自然科学基金项目;编号:2011011015-1;山西省回国人员科研资助项目;编号:20120706ZX

摘　　要：针对静态图像融合方法对于序列图像无法同时满足融合速率及融合质量的问题,提出了一种基于关键帧提取的红外与可见光序列图像快速融合方法。首先对红外和可见光图像序列进行关键帧提取,为体现本文方法的普遍适用性,选取静态图像中融合效果好的算法将两序列中对应的关键帧融合,如小波变换以及主成分分析法。然后对剩余图像帧利用最简单快捷的加权平均算法进行融合。而且相较于现有的只针对于目标进行提取和融合来提高融合速率的序列图像融合方法,在不明确目标的情况下,本文方法对整幅图像进行融合,更加合理且融合结果更适合人眼观察。最后通过对同一场景相同帧数的红外与可见光序列图像进行融合,实验结果表明本文基于关键帧提取的融合方法快速且保证了融合质量。In view of the problem of fusion speed and quality can not be ensured at the same time by static image fusion method,we proposed a fast infrared and visible sequences image fusion method based on key frames extraction.Firstly,extract key frames from the infrared and visible sequences,and fuse the corresponding key frames into those two sequences using the method which has good fusion result in static image.To reflect the universality of this method,we choose WT fusion and PCA fusion as example.Then fuse the rest of the sequences image with the weighted average fusion algorithm.In the case of the target being uncertain,the proposed method fusing the whole image is more reasonable and better for observation while compared with the existing sequences image fusion algorithm which improves speed with only extracting and fusing the target in image.Finally,we fuse the visible and infrared sequences image with the same scene and frame number,and results show that the proposed approach based on key frames extraction ensures the good quality and fast speed at the same time.

关 键 词：序列图像快速融合 可见光图像 红外图像 关键帧提取 

分 类 号：TP732.2[自动化与计算机技术—检测技术与自动化装置]