Domain Fingerprints for No-reference Image Quality Assessment

Weihao Xia1      Yujiu Yang1      Jing-Hao Xue2
1 Tsinghua University       2 UCL

Abstract


Human fingerprints are detailed and nearly unique markers of human identity. Such a unique and stable fingerprint is also left on each acquired image. It can reveal how an image was degraded during the image acquisition procedure and thus is closely related to the quality of an image. In this work, we propose a new no-reference image quality assessment (NR-IQA) approach called domain-aware IQA (DA-IQA), which for the first time introduces the concept of domain fingerprint to the NR-IQA field. The domain fingerprint of an image is learned from image collections of different degradations and then used as the unique characteristics to identify the degradation sources and assess the quality of the image. To this end, we design a new domain-aware architecture, which enables simultaneous determination of both the distortion sources and the quality of an image. With the distortion in an image better characterized, the image quality can be more accurately assessed, as verified by extensive experiments, which show that the proposed DA-IQA performs better than almost all the compared state-of-the-art NR-IQA methods.

Materials


  • arxiv
  •                   

  • github (coming)

Framework



The framework of the proposed DA-NR-IQA. Given a dataset that contains several collections of images with different degradations (a), refer as $n$ domains, we first randomly select a distorted image together with its specific domain label from a certain domain $I_d^i \in \mathcal{D}_{i}$, the generator of Domain-Aware Image Restoration Network (b) aims to separate the distortion $d$ from $I_d$ to get the restoration $I_r$ while the discriminator tries to distinguish if the input image is real or fake and domain discriminator recognizes the category. The original distorted image and its discrepancy map are fed into Discrepancy-Guided Quality Regression Network (c). Features are extracted by a CNN and fused as difference, concatenation with high-level semantic vector. The double arrows means the corresponding module are Siamese Network. Then the fused feature is regressed to a patchwise quality and weight estimation.

Results



Citation

@article{xia2020domain,
  title={Domain Fingerprints for No-reference Image Quality Assessment},
  author={Xia, Weihao and Yang, Yujiu and Xue, Jing-Hao and Xiao, Jing},
  journal={IEEE Transactions on Circuits and Systems for Video Technology (T-CSVT)},
  year={2020},
}