EvReflection: Event-Driven Micro-Dynamics for Reflection Removal

Despite remarkable progress in reflection removal, current methods primarily exploit static image priors from a single frame and still suffer from severe residual artifacts due to the inherent ambiguity between the reflection and transmission layers. In this paper, we propose leveraging event signals to break this ambiguity. By employing event cameras to capture micro-dynamics, we reveal the differential motion between these two layers. We thereby present a novel event-driven reflection removal network, EvReflection, that utilizes these dynamic cues for layer separation. Specifically, we design a Micro-Dynamics Decoupler to disentangle layer-specific motions from event streams as priors, which then guide a Parallax-Attention Rectifier to cleanly remove artifacts from the RGB image. Furthermore, to address data scarcity, we develop a parallax-aware simulation pipeline and construct the EVR$^2$ benchmark dataset, the first real-world dataset for this task. Extensive experiments demonstrate that EvReflection achieves state-of-the-art performance on both synthetic and real-world benchmarks, surpassing the best competing method by more than 1.6 dB and 1.2 dB in PSNR, respectively. The code, dataset, and pre-trained models are available at \href{https://github.com/JiaxiaoWang/EvReflection}{https://github.com/JiaxiaoWang/EvReflection}.

When you take a photo through a window, the glass often creates annoying reflections that blur the background scene you actually want to capture. Current methods try to remove these reflections from a single static photo, but they often fail because it is incredibly difficult for a network to guess which parts of the picture belong to the reflection and which belong to the real background. To solve this, we propose using a novel type of sensor called event camera. Unlike normal cameras, event cameras can record changes in light caused by tiny, micro-movements, such as the natural shaking of a person’s hand. Because the reflection on the glass and the background behind it move in different patterns, these tiny movements provide crucial clues. Our method, EvReflection, uses these movement clues to perfectly separate and peel away the reflection layer, restoring a clear image. We also build the first real-world dataset to test this technology. Our approach significantly outperforms existing methods, opening up new possibilities for improving smartphone photography and autonomous driving in challenging, reflective environments.