基于目标增强的红外与可见光图像融合算法研究  

Research on infrared and visible image fusion algorithmbased on target enhancement

作　　者：丁麒文 祁海军 赵金博 李江勇[1] DING Qi-wen;QI Hai-jun;ZHAO Jin-bo;LI Jiang-yong(North China Research Institute of Electric-Optics,Beijing 100015,China;Beijing Bop Opto-Electronics Technology Co.Ltd.,Beijing 100015,China)

机构地区：[1]华北光电技术研究所,北京100015 [2]北京波谱华光科技有限公司,北京100015

出　　处：《激光与红外》2023年第3期457-463,共7页Laser & Infrared

摘　　要：针对传统多尺度变换融合方法不能有效保留红外图像中热辐射信息及高对比度特征,本文提出了一种基于目标增强多尺度变换的红外和可见光图像融合算法。首先,对可见光图像进行预处理,自适应地提高其对比度。其次,对红外和可见光图像分别使用拉普拉斯金字塔(LP)分解为高频以及低频分量。然后,使用分解后的红外低频信息确定低频段融合权重,并引入参数λ控制融合图像中红外信息比例。最后,使用拉普拉斯金字塔逆变换重构融合图像。实验结果表明,所提出的方法可以生成具有明显突出目标和丰富细节的融合图像,并在主观视觉和客观指标上具有更好的表现。In this paper,a target-enhanced multiscale transform-based fusion algorithm for infrared and visible images is proposed in view of the fact that the traditional multiscale transform fusion methods cannot effectively retain the thermal radiation information and high contrast features in infrared images.Firstly,the visible images are pre-processed to adaptively improve their contrast.Secondly,the IR and visible images are decomposed into high-frequency as well as low-frequency components using Laplace pyramids(LP),respectively.Then,the decomposed infrared low-frequency information is used to determine the low-frequency band fusion weights,and the parameterλis introduced to control the proportion of infrared information in the fused image.Finally,the fused image is reconstructed using the Laplacian pyramid inverse transform.The experimental results show that the proposed method can generate fused images with clearly highlighted targets and rich details,and better performance in terms of subjective visual and objective metrics.

关 键 词：图像融合 多尺度变换 拉普拉斯金字塔 红外图像 可见光图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]