Albedo Estimation via Latent Bridge Matching

2026-09-09

AI Summary

本文通过引入基于潜桥匹配的新架构解决了反照率估计中的物理一致性不足、计算成本高和泛化能力有限的问题。

Abstract

Recent advances in Intrinsic Image Decomposition (IID) have increasingly relied on generative models. However, progress remains limited by three key challenges: (a) insufficient physical consistency, (b) high computational cost at inference time, and (c) limited generalization capabilities. In this work, we show that latent bridge matching (LBM) effectively addresses these limitations for albedo estimation. We introduce a novel LBM-based architecture that enforces physical consistency through a pixel reconstruction loss, benefits from the inherent efficiency of LBM low-cost inference, and improves generalization across diverse datasets by incorporating a shading conditioning. In this extended version, we additionally show that conditioning the shading estimator itself on the predicted albedo further improves reconstruction fidelity, and we benchmark our best model against stateof-the-art IID methods across five real and synthetic datasets.

Problem

• Albedo Estimation
• Intrinsic Image Decomposition
• Physical Consistency
• Computational Cost
• Generalization

Innovation

• Latent Bridge Matching
• Albedo Estimation
• Physical Consistency
• Shading Conditioning
• Low-cost Inference