EvReflection: Event-Driven Micro-Dynamics for Reflection Removal

2026-08-06

AI Summary

This work addresses the limitations of existing reflection removal methods, which rely on single-frame static priors and often fail to effectively disentangle reflection and transmission layers, resulting in noticeable artifacts. To overcome this, the paper introduces event cameras for the first time to capture microsecond-level dynamic cues, exploiting the motion disparity between the two layers. A micro-dynamic decoupler is designed to extract layer-specific motion priors, which guide a disparity-aware attention refinement module for high-fidelity reflection removal. The contributions include EVR²—the first real-world event-image reflection removal dataset—along with a disparity-aware simulation pipeline and an end-to-end learning framework. Experiments demonstrate state-of-the-art performance on both synthetic and real data, with PSNR improvements of 1.6 dB and 1.2 dB over the best existing method, respectively.

Abstract

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 https://github.com/JiaxiaoWang/EvReflection.

Problem

• reflection removal
• event cameras
• layer ambiguity
• residual artifacts
• micro-dynamics

Innovation

• event camera
• reflection removal
• micro-dynamics
• parallax attention
• layer separation