Single Image Reflection Separation via Component Synergy

TL;DR

This paper introduces a novel reflection separation method using a learnable residue model and a dual-stream network with semantic pyramid encoding, achieving superior results on real-world datasets.

Contribution

It proposes a general superposition model with a learnable residue term and a specialized network architecture for improved reflection separation.

Findings

Outperforms state-of-the-art methods on benchmark datasets

Effective residual information capture during decomposition

Robust performance demonstrated through extensive experiments

Abstract

The reflection superposition phenomenon is complex and widely distributed in the real world, which derives various simplified linear and nonlinear formulations of the problem. In this paper, based on the investigation of the weaknesses of existing models, we propose a more general form of the superposition model by introducing a learnable residue term, which can effectively capture residual information during decomposition, guiding the separated layers to be complete. In order to fully capitalize on its advantages, we further design the network structure elaborately, including a novel dual-stream interaction mechanism and a powerful decomposition network with a semantic pyramid encoder. Extensive experiments and ablation studies are conducted to verify our superiority over state-of-the-art approaches on multiple real-world benchmark datasets. Our code is publicly available at https://github.com/mingcv/DSRNet.