多曝光图像融合方法综述  

Review of Multi-Exposure Image Fusion Methods

作　　者：祝新力 张雅声 方宇强 张喜涛 许洁平 罗迪 Zhu Xinli;Zhang Yasheng;Fang Yuqiang;Zhang Xitao;Xu Jieping;Luo Di(Department of Graduate Management,Space Engineering University,Beijing 101416,China;Space Engineering University,Beijing 101416,China)

机构地区：[1]航天工程大学研究生院,北京101416 [2]航天工程大学,北京101416

出　　处：《激光与光电子学进展》2023年第22期15-32,共18页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(61906213)。

摘　　要：高动态范围成像图像是真实表示自然场景中高动态范围亮度的图像,可以反映更多自然场景的信息。多曝光融合以无需改进硬件、算法流程简单的优点成为重建高动态范围图像的重要手段之一,并已在手机相机、工业相机等多个领域得到广泛应用。首先,分别依据融合层次、运动像素处理方式对静态场景、动态场景的多曝光图像融合方法进行分类总结,并对基于深度学习的方法进行单独分析总结。其次,针对多曝光图像融合的相关数据集和性能评价指标进行综述,并对融合方法使用的性能评价指标进行汇总。最后,对多曝光图像融合研究值得关注的问题进行展望,提供了后续相关研究的思路。High dynamic range imaging images are images that truly represent the high dynamic range brightness of natural scenes,and can reflect more information about natural scenes.Multi-exposure fusion has become one of the important means to reconstruct high dynamic range images due to its advantages of no need to improve hardware and simple algorithm process,and has been widely used in mobile phone cameras,industrial cameras,and other fields.In this paper,the multi-exposure image fusion methods for static scenes and dynamic scenes were classified and summarized according to the fusion level and motion pixel processing methods,and the methods based on deep learning were analyzed and summarized separately.Secondly,the relevant datasets and performance evaluation indicators of multi-exposure image fusion were reviewed,and the performance evaluation indicators used in the fusion method were summarized.Finally,the issues worthy of attention in multi-exposure image fusion research were prospected,and ideas for follow-up related research were provided.

关 键 词：多曝光图像融合 高动态范围图像 静态场景 动态场景 深度学习 

分 类 号：O436[机械工程—光学工程]