Summary
Assessing how good a picture looks without comparing it to a perfect original is hard because existing methods focus mostly on what the picture shows, not how clear or detailed it is. The authors noticed that models trained to understand images and words tend to miss fine visual flaws. They created a new approach that helps these models spot subtle image quality problems by focusing on parts of the image features related to imperfections rather than the overall meaning. Their method also improves how images are resized to keep important details for quality checks. Tests show this approach works better than previous ones at judging image quality accurately.
What this means in practice
- •For image processing teams: Enhance automatic image quality scoring in photo editing software to detect subtle visual distortions more accurately.
- •For video streaming engineers: Improve real-time video quality monitoring by better capturing fine details lost in compression and scaling operations.
Authors
Yu Li, Zhengran Shen, Yachun Mi, Puchao Zhou, Shaohui Liu
Abstract
Leveraging Large Vision-Language Models like CLIP has recently set new benchmarks for No-Reference Image Quality Assessment (NR-IQA). However, the contrastive pretraining of CLIP inherently prioritizes semantic invariance, which often suppresses subtle perceptual signals, a phenomenon we term perceptual submergence. Furthermore, standard preprocessing techniques (e.g., cropping and interpolation) further exacerbate the loss of critical high-frequency quality cues. In this paper, we propose the Cross-modal Perception Alignment Adapter (CMPA), a manifold-aware framework designed to disentangle perceptual distortions from dominant semantics. CMPA introduces a Perception-Sensitive Feature Extractor (PFE) that projects CLIP features into a compact, low-dimensional subspace, explicitly magnifying distortion-induced off-manifold deviations. Subsequently, a Cross-Modal Perception Alignment Injector (PAI) aligns these features with quality-aware text anchors and re-injects them into the backbone. To ensure input fidelity, we also devise a Residual-enhanced Perceptual Downscaling strategy that adaptively compensates for resolution-induced information loss using Just Noticeable Difference (JND) guided frequency re-injection. Extensive evaluations on several benchmark datasets demonstrate that our approach significantly outperforms state-of-the-art methods, effectively recovering the perceptual signals submerged in semantic-dense representations.