F2HDR: Two-Stage HDR Video Reconstruction via Flow Adapter and Physical Motion Modeling

TL;DR

F2HDR is a novel two-stage framework for HDR video reconstruction that effectively handles complex motion and exposure variations, producing high-quality, ghost-free HDR videos from alternating-exposure LDR sequences.

Contribution

The paper introduces a flow adapter and physical motion modeling to improve inter-frame alignment and detail restoration in HDR video reconstruction.

Findings

Achieves state-of-the-art results on real-world benchmarks.

Effectively reduces ghosting and noise in dynamic scenes.

Handles large motion and exposure changes robustly.

Abstract

Reconstructing High Dynamic Range (HDR) videos from sequences of alternating-exposure Low Dynamic Range (LDR) frames remains highly challenging, especially under dynamic scenes where cross-exposure inconsistencies and complex motion make inter-frame alignment difficult, leading to ghosting and detail loss. Existing methods often suffer from inaccurate alignment, suboptimal feature aggregation, and degraded reconstruction quality in motion-dominated regions. To address these challenges, we propose F2HDR, a two-stage HDR video reconstruction framework that robustly perceives inter-frame motion and restores fine details in complex dynamic scenarios. The proposed framework integrates a flow adapter that adapts generic optical flow for robust cross-exposure alignment, a physical motion modeling to identify salient motion regions, and a motion-aware refinement network that aggregates complementary information while removing ghosting and noise. Extensive experiments demonstrate that F2HDR achieves state-of-the-art performance on real-world HDR video benchmarks, producing ghost-free and high-fidelity results under large motion and exposure variations.