arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2512.09016cs.CV
收藏

学习去除事件相机中的光斑

Learning to Remove Lens Flare in Event Camera

National University of Singapore(国立新加坡大学);Tsinghua University(清华大学);Institute for Infocomm Research, A*STAR(资讯通信研究院(A*STAR));IPAL, CNRS IRL 2955, Singapore(IPAL,CNRS IRL 2955,新加坡);CerCo, CNRS UMR 5549, Université Toulouse III(CerCo,CNRS UMR 5549,图卢兹第三大学)

Haiqian Han, Lingdong Kong, Jianing Li, Ao Liang, Chengtao Zhu, Jiacheng Lyu, Lai Xing Ng, Xiangyang Ji, Wei Tsang Ooi, Benoit R. Cottereau

AI总结:

本文提出E-Deflare框架,通过物理基础模型和基准数据集,实现了事件相机中光斑的有效去除,提升了视觉系统的性能。

摘要:

事件相机因其高时间分辨率和动态范围有潜力革新视觉系统,但它们仍然容易受到光斑的影响,这是一种基本的光学伪影,会导致严重退化。在事件流中,这种光学伪影形成了一种复杂的时空扭曲,长期以来被忽视。我们提出了E-Deflare,这是首个系统性框架,用于从事件相机数据中去除光斑。我们首先通过推导非线性抑制机制的物理基础前向模型来建立理论基础。这一见解使我们能够创建E-Deflare基准,这是一个包含大规模模拟训练集E-Flare-2.7K和首个-ever配对现实世界测试集E-Flare-R的综合性资源,后者由我们新的光学系统捕获。借助这一基准,我们设计了E-DeflareNet,实现了最先进的恢复性能。广泛的实验验证了我们的方法,并展示了对下游任务的明显好处。代码和数据集已公开可用。

英文摘要:

Event cameras have the potential to revolutionize vision systems with their high temporal resolution and dynamic range, yet they remain susceptible to lens flare, a fundamental optical artifact that causes severe degradation. In event streams, this optical artifact forms a complex, spatio-temporal distortion that has been largely overlooked. We present E-Deflare, the first systematic framework for removing lens flare from event camera data. We first establish the theoretical foundation by deriving a physics-grounded forward model of the non-linear suppression mechanism. This insight enables the creation of the E-Deflare Benchmark, a comprehensive resource featuring a large-scale simulated training set, E-Flare-2.7K, and the first-ever paired real-world test set, E-Flare-R, captured by our novel optical system. Empowered by this benchmark, we design E-DeflareNet, which achieves state-of-the-art restoration performance. Extensive experiments validate our approach and demonstrate clear benefits for downstream tasks. Code and datasets are publicly available.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑