跳到正文
arXiv cs.AI· Mahdi Naseri, Zhou Wang·· 2 小时前AI 评分36

SHAMISA:面向无参考图像质量评估的自监督结构化隐式关联建模

SHAMISA: SHAped Modeling of Implicit Structural Associations for Self-supervised No-Reference Image Quality Assessment

AI 导读

SHAMISA 是一种非对比自监督框架,利用无标注失真图像进行无参考图像质量评估(NR-IQA)预训练,无需主观评分标注和对比损失。其组合式失真引擎从连续参数空间生成失真,保证每次仅一个失真因素变化,实现对表征相似性的细粒度控制。预训练后的编码器冻结并接入线性回归器,在合成、真实及跨数据集 NR-IQA 基准上表现强劲,跨数据集迁移能力有提升。

正文 · 原文

View PDF HTML (experimental)

Abstract:No-Reference Image Quality Assessment (NR-IQA) aims to estimate perceptual quality without access to a pristine reference. Learning an NR-IQA model faces a fundamental bottleneck because it requires many costly human perceptual labels. We propose SHAMISA, a non-contrastive self-supervised framework that learns from unlabeled distorted images through explicitly structured relational supervision. Unlike prior methods that impose rigid binary similarity constraints, SHAMISA shapes soft, controllable implicit structural associations that are both distortion-aware and content-sensitive. A key innovation is its compositional distortion engine, which generates an uncountable family of degradations from continuous parameter spaces and groups the resulting images so that exactly one distortion factor varies at a time. This construction enables fine-grained control over representational similarity during training. Images with shared distortion patterns are pulled together in the embedding space, while severity variations produce structured shifts. Dual-source relation graphs integrate known degradation profiles with emergent structural affinities throughout pre-training. The learned encoder is then frozen and paired with a linear regressor. Extensive experiments on synthetic, authentic, and cross-dataset NR-IQA benchmarks demonstrate strong overall performance, improved transfer in the reported cross-dataset comparisons, and favorable qualitative robustness diagnostics. Self-supervised pre-training requires no IQA opinion-score annotations or contrastive loss.
Comments: Accepted manuscript version. Includes supplementary material
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Cite as: arXiv:2603.13669 [cs.CV]
  (or arXiv:2603.13669v3 [cs.CV] for this version)
  https://doi.org/10.48550/arXiv.2603.13669

arXiv-issued DOI via DataCite

Journal reference: IEEE Transactions on Image Processing 35 (2026) 11216-11227
Related DOI: https://doi.org/10.1109/TIP.2026.3744700

DOI(s) linking to related resources

Submission history

From: Mahdi Naseri [view email]
[v1] Sat, 14 Mar 2026 00:37:26 UTC (7,265 KB)
[v2] Tue, 17 Mar 2026 17:34:31 UTC (7,265 KB)
[v3] Tue, 17 Nov 2026 21:56:49 UTC (7,267 KB)

来源:arXiv cs.AI · arxiv.org