PASDiff: Physics-Aware Semantic Guidance for Joint Real-world Low-Light Face Enhancement and Restoration

TL;DR

PASDiff is a novel physics-aware diffusion model that enhances and restores low-light face images by integrating semantic guidance, facial priors, and photometric constraints, outperforming existing methods.

Contribution

It introduces a training-free, physics-aware diffusion approach with semantic guidance and facial priors for low-light face enhancement and restoration.

Findings

Outperforms existing methods in natural illumination and color recovery

Achieves better identity preservation in low-light face images

Constructs a new benchmark with 700 real-world low-light facial images

Abstract

Face images captured in real-world low light suffer multiple degradations-low illumination, blur, noise, and low visibility, etc. Existing cascaded solutions often suffer from severe error accumulation, while generic joint models lack explicit facial priors and struggle to resolve clear face structures. In this paper, we propose PASDiff, a Physics-Aware Semantic Diffusion with a training-free manner. To achieve a plausible illumination and color distribution, we leverage inverse intensity weighting and Retinex theory to introduce photometric constraints, thereby reliably recovering visibility and natural chromaticity. To faithfully reconstruct facial details, our Style-Agnostic Structural Injection (SASI) extracts structures from an off-the-shelf facial prior while filtering out its intrinsic photometric biases, seamlessly harmonizing identity features with physical constraints. Furthermore, we construct WildDark-Face, a real-world benchmark of 700 low-light facial images with complex degradations. Extensive experiments demonstrate that PASDiff significantly outperforms existing methods, achieving a superior balance among natural illumination, color recovery, and identity consistency.