MoCRA: Mixture of Compositional Rank-1 Atoms for 4K All-in-One Video Restoration

2026-08-03

AI Summary

This work addresses the challenge of jointly restoring real-world 4K videos degraded by multiple concurrent, unlabeled distortions—including haze, rain streaks, low light, and noise—by proposing MoCRA, a unified restoration framework. MoCRA leverages degradation-specific frequency characteristics through a composite conditional mechanism that separately handles downsampling-friendly degradations (e.g., haze, low light) and native-resolution artifacts (e.g., raindrops, noise) across scales. A sparse recombined rank-1 atomic dictionary drives a coarse-to-fine dual-branch architecture for effective recovery. The study introduces UHV-4K-AIO, the first 4K benchmark unifying all four degradation types, and presents a parameter-efficient, optical-flow-free architecture capable of native 4K real-time processing. Experiments demonstrate that MoCRA achieves state-of-the-art average PSNR on the all-in-one 4K restoration task, with inference at 0.5 seconds per frame—over three times faster than the fastest baseline—while maintaining temporal stability comparable to optical-flow-based methods.

Abstract

Real-world video arrives hazy, rainy, dark, or noisy, and a deployable restorer faces three demands at once: no degradation label, native 4K output, and stability in playback. Existing methods answer them separately and break on the joint problem, because per-frame degradation readings flip between frames, downsampled proxies erase the rain and noise they are meant to remove, and dense temporal alignment does not fit 4K memory. No paired benchmark even poses that problem, so we build one. UHV-4K-AIO renders physically modeled haze, rain, sensor noise, and low light over the same 100 clean 4K clips with shared depth and motion, and its construction exposes the split MoCRA is built on: haze and low light survive aggressive downsampling, while rain and noise exist only at native scale. Band-matched compositional conditioning follows, spending conditioning capacity, computation, and supervision in the band where each degradation lives. One dictionary of rank-1 atoms, recomposed sparsely per frame, conditions both a once-per-clip coarse branch and a shallow native-resolution refiner, in 3.6M parameters and with no optical flow. Trained once for all four tasks, MoCRA takes the best task-mean PSNR of eleven retrained image and video baselines, holds warping error at the level of the flow-based video models while never estimating motion, and restores native 4K in under half a second, against 1.7 seconds for the fastest baseline.

Problem

• video restoration
• 4K resolution
• degradation mixture
• label-free
• temporal stability

Innovation

• Mixture of Rank-1 Atoms
• 4K Video Restoration
• All-in-One Degradation Handling
• Band-Matched Conditioning
• Flow-Free Temporal Consistency