Unsupervised Methods for Video Quality Improvement: A Survey of Restoration and Enhancement Techniques

2025-07-11

AI Summary

This paper addresses unsupervised video restoration and enhancement—aiming to improve visual quality and support downstream vision tasks—under the practical constraint of lacking paired ground-truth data. To tackle this challenge, it systematically surveys mainstream approaches grounded in domain adaptation, self-supervised signal design, and blind-spot networks, and introduces, for the first time, a multi-dimensional taxonomy encompassing model architectures, loss functions, and degradation modeling. It further proposes a synthetic-data-driven evaluation paradigm to uniformly benchmark existing methods across denoising, frame interpolation, and super-resolution. Key contributions include: (i) establishing a principled pathway for video quality enhancement without paired supervision; (ii) revealing the synergistic mechanism between noise modeling and structural priors; and (iii) providing a comprehensive theoretical framework and practical guidelines for future research. (149 words)

Technology Category

Computer Vision: Video Understanding & Activity AnalysisMachine Learning: Transfer, Domain Adaptation, Multi-Task LearningSearch and Optimization: Learning to Search

Application Category

Search and Retrieval-Augmented AI: Web evaluation methodologies and metricsUser Modeling, Personalization and Recommendation: Fairness-aware retrieval and rankingGraph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Video restoration and enhancement are critical not only for improving visual quality, but also as essential pre-processing steps to boost the performance of a wide range of downstream computer vision tasks. This survey presents a comprehensive review of video restoration and enhancement techniques with a particular focus on unsupervised approaches. We begin by outlining the most common video degradations and their underlying causes, followed by a review of early conventional and deep learning methods-based, highlighting their strengths and limitations. We then present an in-depth overview of unsupervised methods, categorise by their fundamental approaches, including domain translation, self-supervision signal design and blind spot or noise-based methods. We also provide a categorization of loss functions employed in unsupervised video restoration and enhancement, and discuss the role of paired synthetic datasets in enabling objective evaluation. Finally, we identify key challenges and outline promising directions for future research in this field.

Problem

• Surveying unsupervised video restoration and enhancement techniques
• Addressing common video degradations and their causes
• Exploring challenges and future research directions

Innovation

• Unsupervised video restoration via domain translation
• Self-supervision signal design for enhancement
• Blind spot or noise-based improvement methods