Preference-Guided Debiasing for No-Reference Enhancement Image Quality Assessment

TL;DR

This paper introduces a preference-guided debiasing framework for no-reference enhancement image quality assessment, improving generalization by removing enhancement-specific biases and focusing on perceptual quality cues.

Contribution

It proposes a novel contrastive learning-based embedding space and a debiasing method to enhance the robustness and cross-algorithm generalization of EIQA models.

Findings

Outperforms existing methods on public benchmarks.

Effectively reduces algorithm-induced bias.

Improves robustness and generalization across enhancement algorithms.

Abstract

Current no-reference image quality assessment (NR-IQA) models for enhanced images often struggle to generalize, as they tend to overfit to the distinct patterns of specific enhancement algorithms rather than evaluating genuine perceptual quality. To address this issue, we propose a preference-guided debiasing framework for no-reference enhancement image quality assessment (EIQA). Specifically, we first learn a continuous enhancement-preference embedding space using supervised contrastive learning, where images generated by similar enhancement styles are encouraged to have closer representations. Based on this, we further estimate the enhancement-induced nuisance component contained in the raw quality representation and remove it before quality regression. In this way, the model is guided to focus on algorithm-invariant perceptual quality cues instead of enhancement-specific visual fingerprints. To facilitate stable optimization, we adopt a two-stage training strategy that first learns the enhancement-preference space and then performs debiased quality prediction. Extensive experiments on public EIQA benchmarks demonstrate that the proposed method effectively mitigates algorithm-induced representation bias and achieves superior robustness and cross-algorithm generalization compared with existing approaches.