PhaSR：物理对齐先验的泛化图像阴影去除
PhaSR: Generalized Image Shadow Removal with Physically Aligned Priors
把 Gray-world 归一化、对数域 Retinex 分解与动态范围重组整合成闭式光照校正模块，并用扩展的差分注意力对齐深度几何与 DINO-v2 语义特征，可为其他光照解耦类复原任务（低光照增强、去雾）提供可复用的先验对齐范式。
原文依据 Geometric-Semantic Rectification Attention (GSRA) extends differential attention to cross-modal alignment
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论在多样光照条件下去除阴影需要将光照与固有反射率解耦——当物理先验未被恰当对齐时，这一挑战会进一步加剧。我们提出 PhaSR（物理对齐阴影去除），通过双层先验对齐来解决该问题，使其从单光源阴影到多源环境光照均能保持稳健性能。首先，物理对齐归一化（PAN）通过 Gray-world 归一化、对数域 Retinex 分解与动态范围重组执行闭式光照校正，抑制色度偏差。其次，几何-语义矫正注意力（GSRA）将差分注意力扩展到跨模态对齐，协调深度导出的几何信息与 DINO-v2 语义嵌入，以解决变化光照下的模态冲突。实验表明，该方法在阴影去除上具有竞争力，复杂度更低，并能泛化到传统方法在多源光照下失效的环境光照场景。
英文原文摘要 1014 字符
Shadow removal under diverse lighting conditions requires disentangling illumination from intrinsic reflectance--a challenge compounded when physical priors are not properly aligned. We propose PhaSR (Physically Aligned Shadow Removal), addressing this through dual-level prior alignment to enable robust performance from single-light shadows to multi-source ambient lighting. First, Physically Aligned Normalization (PAN) performs closed-form illumination correction via Gray-world normalization, log-domain Retinex decomposition, and dynamic range recombination, suppressing chromatic bias. Second, Geometric-Semantic Rectification Attention (GSRA) extends differential attention to cross-modal alignment, harmonizing depth-derived geometry with DINO-v2 semantic embeddings to resolve modal conflicts under varying illumination. Experiments show competitive performance in shadow removal with lower complexity and generalization to ambient lighting where traditional methods fail under multi-source illumination.