Debiased Mapping for Full-Reference Image Quality Assessment

Baoliang Chen, Hanwei Zhu, Lingyu Zhu, Shanshe Wang, Jingshan Pan, Shiqi Wang*

*Corresponding author for this work

Research output: Journal Publications and ReviewsRGC 21 - Publication in refereed journalpeer-review

Abstract

An ideal full-reference image quality (FR-IQA) model should exhibit both high separability for images with different quality and compactness for images with the same or indistinguishable quality. However, existing learning-based FR-IQA models that directly compare images in deep-feature space, usually overly emphasize the quality separability, neglecting to maintain the compactness when images are of similar quality. In our work, we identify that the perception bias mainly stems from an inappropriate subspace where images are projected and compared. For this issue, we propose a Debiased Mapping based quality Measure (DMM), leveraging orthonormal bases formed by singular value decomposition (SVD) in the deep features domain. The SVD effectively decomposes the quality variations into singular values and mapping bases, enabling quality inference with more reliable feature difference measures. Extensive experimental results reveal that our proposed measure could mitigate the perception bias effectively and demonstrates excellent quality prediction performance on various IQA datasets. Codes are available at https://github.com/Baoliang93/DMM. © 2025 IEEE.
Original languageEnglish
JournalIEEE Transactions on Multimedia
DOIs
Publication statusOnline published - 13 Feb 2025

Funding

This work was supported in part by the National Key Research and Development Project (2022YFF0902203), in part by the National Natural Science Foundation of China under Grant 62401214, in part by the New 20 Project of Higher Education of Jinan (No. 202228024), in part by ITF Project GHP/044/21SZ, in part by RGC General Research Fund 11203220/11200323.

Research Keywords

  • deep feature
  • Image quality assessment
  • natural scene statistics
  • nuclear normal
  • weight distribution

Fingerprint

Dive into the research topics of 'Debiased Mapping for Full-Reference Image Quality Assessment'. Together they form a unique fingerprint.

Cite this