基于多尺度分解的双曝光图像融合方法  

Two-exposure image fusion method based on multi-scale decomposition

作　　者：田浩南[1] 张美君[1] 卜和阳[1] Tian Haonan;Zhang Meijun;Bu Heyang(Changchun Institute of Optics,Fine Mechanics and Physics,Chinese Academy of Sciences,Changchun 130033,China)

机构地区：[1]中国科学院长春光学精密机械与物理研究所,长春130033

出　　处：《国外电子测量技术》2024年第9期16-25,共10页Foreign Electronic Measurement Technology

基　　金：国家自然科学基金委员会(62105326)项目资助。

摘　　要：为了有效地提高可见光成像设备的动态范围,降低融合图像的质量对源图像数量的依赖,提出了一种基于多尺度分解的双曝光图像融合算法。该方法只需要一组欠曝光和过曝光图像作为源图像,通过曝光融合即可得到一幅包含丰富信息的图像。首先,依据欠曝光图像和过曝光图像自身的特点,分别进行了自适应曝光调整,充分挖掘图像中潜在的细节信息。然后,提取图像序列的边缘强度、曝光适宜度和色彩饱和度作为评价指标,进而构建出融合权重图。最后,通过金字塔多尺度分解和加权融合得到融合图像。实验选取了15组图像序列,分别从主观和客观两个方面与4种具有代表性的算法进行了对比。实验结果表明,本文算法相比于其他算法,图像质量综合提升了4.9%,具有更强的细节信息保留能力。To effectively improve the dynamic range of visible light imaging devices and reduce the dependency of fused image quality on the number of source images,a dual-exposure image fusion algorithm based on multi-scale decomposition is proposed.This method requires only one set of underexposed and overexposed images as source images to obtain a richly informative image through exposure fusion.Firstly,adaptive exposure adjustment is performed according to the characteristics of the underexposed and overexposed images,fully exploring the potential details in the images.Then,edge intensity,exposure suitability,and color saturation of the image sequence are extracted as evaluation indicators to construct a fusion weight map.Finally,the fused image is obtained through pyramid multi-scale decomposition and weighted fusion.Fifteen groups of image sequences were selected for experiments,and comparisons were made with four representative algorithms from both subjective and objective perspectives.The experimental results show that compared to other algorithms,the proposed algorithm improves overall image quality by 4.9%and has a stronger ability to retain detailed information.

关 键 词：双曝光图像 多尺度分解 图像金字塔 加权融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TN911.73[自动化与计算机技术—计算机科学与技术]