$\text{F}^2\text{HDR}$: Two-Stage HDR Video Reconstruction via Flow Adapter and Physical Motion Modeling

2026-03-16

AI Summary

This work addresses the challenge of ghosting artifacts and detail loss in high dynamic range (HDR) video reconstruction from alternately exposed low dynamic range (LDR) frames under dynamic scenes, where exposure inconsistency and complex motion degrade reconstruction quality. To tackle this, the authors propose a two-stage framework: first, a flow adapter enables robust cross-exposure alignment by integrating physical motion modeling to accurately identify regions with significant motion; second, a motion-aware refinement network adaptively fuses complementary information from multiple frames to suppress ghosting and noise. Evaluated on real-world HDR video benchmarks, the method achieves state-of-the-art performance, demonstrating substantially improved robustness and detail fidelity under large motions and strong exposure variations.

Technology Category

• Computer Vision: Motion & Tracking
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Adversarial Learning & Robustness

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments

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 $\text{F}^2\text{HDR}$, 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 $\text{F}^2\text{HDR}$ 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.

Problem

• HDR video reconstruction
• cross-exposure inconsistency
• complex motion
• ghosting
• inter-frame alignment

Innovation

• HDR video reconstruction
• optical flow adaptation
• physical motion modeling
• ghosting removal
• two-stage framework