MR-IQA: A Unified Margin View of Regression and Ranking for Blind Image Quality Assessment

2026-06-29

AI Summary

This work addresses the lack of a unified theoretical foundation between regression and learning-to-rank paradigms in blind image quality assessment (BIQA), which hinders their joint optimization. The authors propose MR-IQA, a novel framework that, for the first time, reveals a common optimization basis—termed “quality margin”—at the objective level. Building upon this insight, they develop a unified reinforcement learning approach that explicitly models image quality structure by employing pairwise quality margin errors as policy rewards. By integrating regression supervision with preference probability modeling, the proposed method consistently outperforms pure regression- or ranking-based reinforcement learning approaches across six BIQA benchmarks, achieving state-of-the-art average PLCC and SRCC performance.

Abstract

Blind image quality assessment (BIQA) is commonly built on two basic learning paradigms: regression and ranking. Regression calibrates absolute scores, whereas ranking recovers quality structure from ordinal relations. Although joint regression-ranking supervision often improves BIQA, the relation between the two paradigms remains largely empirical and underexplored. In this work, we revisit what underlies regression and ranking and identify pairwise relational distance, termed quality margin, as their common bridge. Our derivation shows that, at the objective-optimization level, both paradigms fit quality margins: regression fits margins induced by score endpoints, while ranking fits transformed or sign-level margins through preference probabilities. Motivated by this insight, we propose MR-IQA, a direct quality-margin optimization framework for reinforcement learning (RL)-based BIQA. MR-IQA samples quality scores and optimizes pairwise margin errors as policy rewards, thereby modeling quality structure more explicitly. Experiments on six BIQA benchmarks show competitive general performance, and controlled comparisons demonstrate that MR-IQA achieves the strongest average PLCC/SRCC over regression- or ranking-based RL methods. Our findings provide a new insight into unifying regression and ranking, offering a theoretical basis for understanding quality-structure modeling in BIQA and beyond.

Problem

• blind image quality assessment
• regression
• ranking
• quality margin
• quality-structure modeling

Innovation

• quality margin
• blind image quality assessment
• regression-ranking unification
• reinforcement learning
• pairwise relational distance