This paper presents a comprehensive survey of low-light image and video enhancement. We begin with the challenging mixed over-/under-exposed images, which are under-performed by existing methods. To this end, we propose two variants of the SICE dataset named SICE\_Grad and SICE\_Mix. Next, we introduce Night Wenzhou, a large-scale, high-resolution video dataset, to address the lack of low-light video datasets that discourages the use of low-light image enhancement (LLIE) methods in videos. Our Night Wenzhou dataset is challenging since it consists of fast-moving aerial scenes and streetscapes with varying illuminations and degradation. We then construct a hierarchical taxonomy, conduct extensive key technique analysis, and performs experimental comparisons for representative LLIE approaches using our proposed datasets and the current benchmark datasets. Finally, we identify emerging applications, address unresolved challenges, and propose future research topics for the LLIE community. Our datasets are available at https://github.com/ShenZheng2000/LLIE_Survey.
翻译：本文介绍了对低光图像和视频增强的综合调查。 我们首先介绍了挑战性混合的超光/低光图像,这些图像在现有方法下表现不足。 为此,我们提出了SICE数据集的两个变体,名为SICE ⁇ Grad和SICE ⁇ Mix。接下来,我们介绍了一个大型高分辨率视频数据集“夜文州”,以解决在视频中缺少低光视频数据集的问题,这不利于使用低光图像增强方法。我们的夜文州数据集具有挑战性,因为它包含快速移动的空中场景和街道景,具有不同的照明和退化作用。我们随后构建了一个等级分类学,进行广泛的关键技术分析,并利用我们提议的数据集和当前的基准数据集对具有代表性的LIEE方法进行实验性比较。最后,我们确定了新出现的应用,解决尚未解决的挑战,并为LIEE社区提出未来的研究课题。我们的数据集可在https://github.com/Sheng2000/LIE_Survey查阅。
相关内容
微信扫码咨询专知VIP会员