Reflection Separation from a Single Image via Joint Latent Diffusion

2026-06-02

AI Summary

This work addresses the challenging problem of separating reflection and transmission layers from a single image under extreme conditions such as strong glare or weak reflections, where existing methods often fail due to insufficient information to recover both layers simultaneously. To overcome this limitation, the authors propose a unified framework based on fine-tuned latent diffusion models, incorporating cross-layer self-attention mechanisms to enhance feature disentanglement. The approach further integrates a decoupled iterative sampling strategy and a learnable image composition function, enabling optimization in the latent space guided by generative priors. This design significantly reduces interference between layers and achieves superior performance over state-of-the-art methods across multiple real-world benchmarks, yielding more robust and high-quality reflection separation results.

Abstract

Single-image reflection separation is highly challenging under extreme conditions like glare or weak reflections. Existing methods often struggle to recover both layers in glare or weak-reflection scenarios because of insufficient information. This paper presents a diffusion model explicitly fine-tuned for this task, leveraging generative diffusion priors for robust separation. Our method simultaneously generates transmission and reflection layers through a unified diffusion model, incorporating a novel cross-layer self-attention mechanism for better feature disentanglement. We further introduce a disjoint sampling strategy to iteratively reduce interference between the layers during diffusion and a latent optimization step with a learned composition function for improved results in complex real-world scenarios. Extensive experiments demonstrate that our approach surpasses state-of-the-art methods on multiple real-world benchmarks. Project page: https://brian90709.github.io/diff-reflection-separation/

Problem

• reflection separation
• single image
• glare
• weak reflection
• layer recovery

Innovation

• diffusion model
• reflection separation
• cross-layer self-attention
• disjoint sampling
• latent optimization