A Lightweight Deep Exclusion Unfolding Network for Single Image Reflection Removal

2025-03-03

AI Summary

Single-image reflection removal (SIRR) aims to decompose a single composite image into its underlying transmission and reflection layers, with the core challenge lying in modeling their mutually exclusive physical priors. This paper proposes an interpretable deep-unfolding network based on mutual exclusivity regularization. It is the first to parameterize the iterative sparse optimization and auxiliary feature update process as a lightweight convolutional network, explicitly embedding inter-layer mutual exclusivity within a model-driven framework. Deep unfolding bridges interpretability and performance by unrolling the optimization steps into learnable network layers. Our method achieves state-of-the-art results on four benchmark datasets, with significant improvements in PSNR and SSIM. Notably, it reduces parameter count to only 8% of the current best-performing method, achieving superior efficiency while preserving high visual fidelity.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Search and Optimization: Non-convex Optimization
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Single Image Reflection Removal (SIRR) is a canonical blind source separation problem and refers to the issue of separating a reflection-contaminated image into a transmission and a reflection image. The core challenge lies in minimizing the commonalities among different sources. Existing deep learning approaches either neglect the significance of feature interactions or rely on heuristically designed architectures. In this paper, we propose a novel Deep Exclusion unfolding Network (DExNet), a lightweight, interpretable, and effective network architecture for SIRR. DExNet is principally constructed by unfolding and parameterizing a simple iterative Sparse and Auxiliary Feature Update (i-SAFU) algorithm, which is specifically designed to solve a new model-based SIRR optimization formulation incorporating a general exclusion prior. This general exclusion prior enables the unfolded SAFU module to inherently identify and penalize commonalities between the transmission and reflection features, ensuring more accurate separation. The principled design of DExNet not only enhances its interpretability but also significantly improves its performance. Comprehensive experiments on four benchmark datasets demonstrate that DExNet achieves state-of-the-art visual and quantitative results while utilizing only approximately 8% of the parameters required by leading methods.

Problem

• Separates reflection-contaminated images into transmission and reflection components.
• Minimizes commonalities between transmission and reflection features.
• Proposes lightweight, interpretable network for single image reflection removal.

Innovation

• Lightweight Deep Exclusion Unfolding Network (DExNet)
• Sparse and Auxiliary Feature Update (i-SAFU) algorithm
• General exclusion prior for feature separation